Effect of Pavement Thicknesson Superpave Mix
Permeability and Density
SPR# 0092-02-14c
margorP hcraeseR y a
w hgiH nis nocs i
W WHRP 05-05
James CrovettiMarquette University
Robert Schmitt University of Wisconsin-Platteville
April 2005
Jeffrey Russell, Hussain U. Bahia, Kunnawee KanitpongUniversity of Wisconsin-Madison
EFFECT OF PAVEMENT THICKNESS ON SUPERPAVE MIX PERMEABILITY AND DENSITY
WisDOT Highway Research Study 0092-02-14
By
Jeffrey Russell, Professor Hussain U. Bahia, Associate Professor
Kunnawee Kanitpong, Research Assistant University of Wisconsin – Madison
Department of Civil and Environmental Engineering 1415 Engineering Drive, Madison, WI 53706-1490
&
Robert Schmitt University of Wisconsin- Platteville
Department of Civil & Environmental Engineering University Plaza, Platteville, WI 53818
&
James Crovetti Marquette University
Department of Civil & Environmental Engineering Haggerty Engineering Hall
Milwaukee, WI 53201
Submitted to
Wisconsin Department of Transportation Division of Transportation Infrastructure Development
Research Coordination Section 4802 Sheboygan Ave., Box 7065, Madison, WI 53707-7910
April 2005
ii
Disclaimer This research was funded through the Wisconsin Highway Research Program by the Wisconsin Department of Transportation and the Federal Highway Administration under Project # 0092-02-14. The contents of this report reflect the views of the authors who are responsible for the facts and the accuracy of the data presented herein. The contents do not necessarily reflect the official views of the Wisconsin Department of Transportation or the Federal Highway Administration at the time of publication.
This document is disseminated under the sponsorship of the Department of Transportation in the interest of information exchange. The United States Government assumes no liability for its contents or use thereof. This report does not constitute a standard, specification or regulation.
The United States Government does not endorse products or manufacturers. Trade and manufacturers’ names appear in this report only because they are considered essential to the object of the document.
iii
Acknowledgement
There are many people who contributed to the completion of this project. Some
of the more important contributors include the members of the Wisconsin Highway
Research Program Flexible Pavement Technical Oversight Committee. Their feedback
and support, especially Erv Dukatz, Judie Ryan, and Tom Brokaw are gratefully
acknowledged.
The authors gratefully acknowledge the support of the Payne and Dolan Inc., Ms.
Signe Reichelt, and the Mathy Construction, Dr. Erv Dukatz for their support in
providing information of the field projects selected in this study. Authors would also like
to thank Mr. Jacques Menard from Marquette University, Mr. Anthony Stakson and Mr.
Ahmed Faheem from the University of Wisconsin-Madison for their assistances in
collecting the field data, and Ms. Susan Brunsell for her coordination in the project.
The authors would also like to thank Mr. Greg Waidley for his support in
finishing the final version of this report and for his review of the document.
iv
Technical Report Documentation Page
1. Report No.
2. Government Accession No
3. Recipient’s Catalog No
4. Title and Subtitle Effect of Pavement Thickness on Superpave Mix Permeability and Density
5. Report Date : April 2005 6. Performing Organization Code 0092-02-14
7. Authors J.S. Russell, Professor, H.U. Bahia, Associate Professor, and K. Kanitpong, Research Assistant
8. Performing Organization Report No.
9. Performing Organization Name and Address University of Wisconsin – Madison Department of Civil and Environmental Engineering 1415 Engineering Drive Madison, WI 53706-2507
10. Work Unit No. (TRAIS) 11. Contract or Grant No. WisDOT SPR# 0092-02-14
12. Sponsoring Agency Name and Address Wisconsin Department of Transportation Division of Transportation Infrastructure Development Research Coordination Section 4802 Sheboygan Ave., Box 7065, Madison, WI 53707-7910
13. Type of Report and Period Covered 14. Sponsoring Agency Code
15. Supplementary Notes 16. Abstract: This research study was conducted to determine the influence of maximum aggregate size, lift thickness, and aggregate source on the density and permeability of asphalt mixtures designed according to the Superpave criteria. The guidelines for the selection of pavement layer thickness based on nominal maximum aggregate size and gradation for use in Wisconsin were developed, and the permeability and density criteria for Superpave mixture designs in Wisconsin based on traffic, lift thickness, field drainage and moisture conditions were recommended. In addition, the laboratory and field permeability testing procedures and equipment for design and quality control of Superpave mixtures in Wisconsin were recommended. This project presents the results of 16 mixes used on 9 field projects, including all critical variables affecting the density and permeability of HMA. The in-place density and field permeability were measured by using the nuclear gauge and the NCAT device, respectively. Field cores were taken for measuring permeability in the laboratory by using the ASTM D5084 method; and laboratory compaction was used to prepare and test samples from loose mixtures recovered from the field. The results from field study indicate that that density and permeability of Superpave mixes are based on project-specific variables. Base type, source, gradation, and Ndes level all influence field density and permeability. For fine-graded mixes, the t/NMAS ratio showed an influence on achieving density, particularly below a ratio of 2 for gravel-source mixes and a ratio of 3 for limestone-source mixes. No clear relationship was found between t/NMAS ratios and permeability. For coarse-graded mixes, mixes compacted at smaller t/NMAS ratios for limestone-source were more permeable than higher ratios, but no trend was observed for the gravel-source mix. It was also found that there is a good correlation between the gradation of aggregate and permeability. As the ratio of (%P1/2 - %P3/8) / (%PNo.4-%PNo.8) increases, the permeability decreases, and as the gaps between the coarse aggregates (%P1/2” and %P3/8”) and/or the fine aggregates (%P4 and %P8) increase, the permeability increases. This could be the effect of differences in aggregate sizes on the internal void structure, and thus measured permeability, of the compacted material. This trend could be used in mix design by controlling the ratio to limit permeability by either reducing the difference between the coarse sieves, fine sieves, or both. In laboratory study, two compaction procedures, called Method A and Method B were used to produce Superpave Gyratory Compacted (SGC) specimens that have similar thickness, air voids, and aggregate orientation of the field cores. The result indicates that Method B, which is based on using Ndesign gyrations for different sample sizes, can be used to produce samples that give permeability values similar to values measured for field cores. The results indicate a good relationship between field permeability (using the NCAT device) and lab permeability measured on field cores of fine-graded mixes with amount of passing No. 8 sieve (P8) higher than 45%. However, the relationship between field
v
permeability and lab permeability measured on field cores of coarse-graded mix (P8 lower than 40%) is very poor. It is therefore concluded that the NCAT permeability device could possibly be used in the field for fine-graded mix (with P8 higher than 45%) to measure a permeability index that is related to the true permeability of field cores as measured by the ASTM D5084. However, to measure the field permeability of coarse-graded mix (P8 lower than 40%), an approach to prevent water leakage along the sealant due to rough pavement surface should be established. For coarse graded mixtures, there appears to be no current alternative better than taking field cores and testing them in the laboratory. For estimating permeability during mixture design, a simple method for preparing and testing permeability of SGC specimens and interpolating based on expected field density is introduced. The results represent a good estimate of the expected in-place field permeability. The recommendations from this study include no changes in the selection of pavement thickness and t/NMAS ratios in the specifications. However, this recommendation does not ensure achieving density nor limit permeability. It is also recommended that for the permeability and density criteria for Superpave mix designs, the target permeability and density values should be developed from in-service pavements with recorded performance histories. For further study, the warranty projects with proven record of performance can be used to define target density and permeability criteria for HMA pavement in Wisconsin. 17. Key Words Density, Permeability, HMA, Superpave, Lift Thickness, NMAS, t/NMAS, Gradation
18. Distribution Statement
No restriction. This document is available to the public through the National Technical Information Service 5285 Port Royal Road Springfield VA 22161
19. Security Classif.(of this report) Unclassified
19. Security Classif. (of this page) Unclassified
20. No. of Pages
21. Price
Form DOT F 1700.7 (8-72) Reproduction of completed page authorized
vi
Executive Summary
Project Summary
The objectives of this study were to determine the influence of maximum
aggregate size, lift thickness, and aggregate source on the density and permeability of
asphalt mixtures designed according to the Superpave criteria. The guidelines for the
selection of pavement layer thickness based on nominal maximum aggregate size and
gradation for use in Wisconsin were developed as well as the permeability and density
criteria for Superpave mixture designs in Wisconsin based on traffic, lift thickness, field
drainage and moisture conditions were recommended. In addition, the effect of void
characteristics, arrangement, and interconnectivity on permeability was evaluated. For
laboratory study, the laboratory and field permeability testing procedures and equipment
for design and quality control of Superpave mixtures in Wisconsin were recommended.
Background
The permeability of asphalt mixtures is well known as a function of aggregate
gradation, density achieved, and distribution of air voids. According to the Superpave
mix design procedure, the gradations on the coarse side of the maximum density line are
being widely used, and these gradations are claimed to be more permeable. Specific
questions were raised whether this trend is due to changes in the air voids distribution, the
lower densities being achieved, or both. Recent studies have also shown that
permeability is a directional property such that orientation of the aggregates, which is
affected by lift thickness and level of compaction, has a significant effect on total
permeability.
vii
The lift thicknesses that Wisconsin has traditionally used are based on the
traditional rule that lift thickness be twice the maximum aggregate size. Since Wisconsin
has decided to move from Marshall design to Superpave mixture design in 2000,
Superpave mixes was found to be harder to compact. Additionally, Superpave guidelines
recommend the lift thickness be a minimum of 3 times the nominal maximum aggregate
size. Accordingly, two problems for Wisconsin were encountered which are that 1) the
current design criteria for overlay thickness will result in thin-lifts of Superpave mixes
that the AASHTO Lead States Committee has reported as having problems with
pavement permeability and achieving pavement density, and 2) these mixes may be
impossible to compact in the field contributing to the permeability problem, even though
they meet laboratory density criteria.
Therefore, a study to evaluate the potential problems and to establish procedures
to relate laboratory density to field study and to estimate or measure permeability during
mixture design is necessary. In addition, the study also needs to define the relationship
between lift thickness and aggregate gradations that will minimize the densification
problem and address the permeability concerns.
Process
To accomplish the objectives of this study, the critical variables that affect the
density and permeability of HMA were initially defined and used in the experimental
design. The research team, in collaboration with the Wisconsin DOT and the
representative of the asphalt paving industry, then selected HMA plants with consistent
aggregate sources. The major aggregate sources representing the most widely used
aggregates in Wisconsin pavements were selected. Other critical variables such as
viii
gradation and nominal maximum aggregate size were also considered in the selection of
Superpave mix and materials used in the study.
A set of projects were selected that allow measuring the effect of different
variables identified in the experimental results. These projects include Superpave
mixtures with different nominal maximum aggregate size, gradations, aggregate sources,
lift thickness and sub-surface layers. The selection was based on a review of WisDOT
projects and other projects that the asphalt industry is involved in. The characteristics of
each project were documented first, and based on specific criteria; the projects were
ranked and matched with required factors to be studied. The highest ranked projects
were selected and reviewed with the members of the flexible pavement TOC to finalize
the list and contact the contractors involved.
In the field study, the in-place densities were measured by using nuclear gauges,
and the field permeability was performed immediately after the density was measured.
The field permeability was measured by using a falling-head permeameter similar to
NCAT device. The field cores were then taken to laboratory after the permeability was
completed. The loose mix from each project was taken to the lab for producing the
laboratory compacted specimens. In the laboratory study, the Superpave gyratory
compactor (SGC) was used to compact the specimens from loose mixes at the same
density as the field cores. The lab permeability was then measured for field cores and
lab-compacted specimens. The relationships between field permeability, lab permeability
of field cores, and lab permeability of lab-compacted specimens were determined from
the results obtained.
ix
Findings and Conclusions
Field Study
It was found that density and permeability of Superpave mixes are based on
project-specific variables. Base type, source, gradation, and Ndes level all influence field
density and permeability. Layer thickness was a factor on a project-specific basis, with
some projects indicating it was significant, while others found it not significant.
For fine-graded mixes, the t/NMAS ratio showed an influence on achieving
density, particularly below a ratio of 2 for gravel-source mixes and a ratio of 3 for
limestone-source mixes. For limestone-source mixes outside the current WisDOT
t/NMAS range of 3 to 5, it was more difficult to achieve density below a ratio of 3, and
possible to achieve a 92% density above a ratio of 5. However, no clear relationship was
found between t/NMAS ratios and permeability.
For coarse-graded mixes, mixes compacted at smaller t/NMAS ratios for
limestone-source were more permeable than higher ratios, but no trend was observed for
the gravel-source mix.
It was also found that gradation of the aggregate could be linked to permeability.
The ratio of (%P1/2 - %P3/8) / (%PNo.4-%PNo.8) had a good correlation with
permeability with high ratios showing lower permeability. In addition, higher
permeability was measured as the gaps increase between the coarse aggregates (%P1/2”
and %P3/8”) and/or the fine aggregates (%P4 and %P8). This suggests that relative
differences in these sieves may have an effect on internal void structure, and thus
measured permeability, of the compacted material. This trend could be used in mix
x
design by controlling the ratio to limit permeability by either reducing the difference
between the coarse sieves, fine sieves, or both.
Laboratory Study
The NCAT field permeability device was found to give results that sometimes
compares well to laboratory measurements done on field cores but not always. For fine-
graded mixture with P8 higher than 45%, field permeability measured by the NCAT
device strongly correlates to laboratory permeability measured on field cores taken from
same pavements section. However, the relationship is not one to one ratio. The field
permeability values could be approximately an order of magnitude higher than the lab
permeability. This could be explained by the multiple flow directions in the field
permeability measurement. The coefficients of correlation for the mathematical
relationship found is high (R2 = 0.80). This indicates that the NCAT permeability
devices, with all its limitations, could be used in the field for fine-graded mixture (with
P8 > 45%) to measure an index of permeability reliability. The measured values can then
be related to true permeability of field cores measured by the ASTM D5084 conducted
under well-controlled conditions. There is a concern, however, in using the NCAT
device for measuring the field permeability of mixtures with P8 lower than 40%, since
very poor correlation was found for the relationship between field and lab permeability in
this study. The modification of NCAT device is therefore necessary in order to prevent
water leakage along the sealant due to rough pavement surface, particularly for mixtures
with coarse gradation.
A method was proposed to compact specimens in the SGC at various sample sizes
that could be used to estimate relatively well the permeability of the specimens taken out
xi
from pavements in the field. The permeability measured on these SGC specimens
correlates to the permeability measured on the field cores with a relationship of one to
one. Therefore, this method (called in the report Method B) could be used for predicting
the permeability of asphalt mixtures in the field. If this method can be validated, then the
permeability can be included as a design requirement.
A method, and related equipment, were developed for quantifying the preferential
void pathways in compacted asphalt layers. The degree of vertically connected void
pathways was found to be best correlated to the pavement layer thickness, with greater
thicknesses producing a reduction in preferential vertical void pathways. Correlations
between field/lab water permeability ratios and preferential vertical void pathways
indicate that field and laboratory permeability values can only be expected to be in near
agreement when the degree of preferential vertical void pathways exceeds 80% for fine
mixes. For coarse mixes with a high degree of preferential vertical void pathways,
field/lab water permeability ratios of 10 or more may be expected.
Recommendations
For the selection of pavement thickness, it is recommended that no changes be
made to the current layer thickness values and t/NMAS ratios in the specifications.
Density and permeability characteristics of Superpave mixes are found to depend on
several project-specific variables, such as base type, source, gradation, Ndes level, layer
thickness, and t/NMAS ratio. No compelling evidence is found in the data to alter layer
thickness and t/NMAS ratios, without accounting for the other remaining project-specific
variables. It is however important to recognize that the current recommendations do not
xii
ensure achieving density nor limit permeability. Difficulty in achieving density or
exceeding acceptable permeability is influenced by several interacting factors.
For the permeability and density criteria for Superpave mix designs, it is
recommended that target permeability and density values ultimately be established from
in-service pavements with recorded performance histories. One such group of pavements
includes accepted warranty projects that have been in service for 5 or more years. Field
permeability and density measures on these pavements can aid in the development of
acceptance values that correlate to good performance.
Until a performance-based determination is made, an interim approach is
recommended that establishes the minimum acceptable density based on median
permeability values. Based on research data included in this report for fine-graded
Superpave mixes, a specified minimum density of 93.8% would be required to limit
permeability to 150x10-5 cm/sec. For coarse-graded Superpave mixes, the research data
does not support the establishment of minimum acceptable densities to control
permeability because of the lack of a unified relationship between density and
permeability that is independent of source or gradation of mixtures. The limit should
remain at 150x10-5 cm/sec but should be measured directly on a core recovered from
pavement section.
xiii
TABLE OF CONTENTS
CHAPTER ONE: INTRODUCTION..............................................................................4 1.1 Background and Problem Statement.............................................................................4 1.2 Literature Review..........................................................................................................6 1.3 Research Objectives....................................................................................................23 1.4 Research Methodology ...............................................................................................24 1.5 Experimental Design...................................................................................................27 1.6 Summary .....................................................................................................................28
CHAPTER TWO: FIELD STUDY ................................................................................30 2.1 Introduction.................................................................................................................30 2.2 Equipment and Methods .............................................................................................33 2.3 Statistical Analysis of Field Studies ...........................................................................36
2.3.1 Fine Mixes ...................................................................................................41 2.3.2 Coarse Mixes ...............................................................................................53 2.3.3 Density Growth............................................................................................58
2.4 Investigation of Specification Criteria ........................................................................64 2.6 Summary of Findings from Field Study .....................................................................70
CHAPTER THREE: LABORATORY DATA ANALYSIS AND DISCUSSIONS ...73 3.1 Introduction.................................................................................................................73 3.2 Field Cores Permeability Testing................................................................................73
3.2.1 Equipment and Methods ..............................................................................73 3.2.2 Density and Permeability Results ................................................................80
3.3 Laboratory Compacted Specimen Testing..................................................................82 3.3.1 Alternatives for Laboratory Compaction Method........................................82
3.3.2 Proposed Compaction Procedure.................................................................85 3.3.3 Density and Permeability Results ................................................................88 3.4 Correlations of Lab and Field Results ........................................................................90 3.4.1 Correlation between Field Density and Lab Density ..................................90
3.4.2 Correlation between Field Permeability and Lab Permeability of Field Cores ..............................................................................................91
3.4.3 Correlation between Laboratory Permeability of Field Cores and Predicted Permeability Using Lab Compacted Specimens...................96 3.5 Summary of Findings of Laboratory Study ................................................................97
CHAPTER FOUR: AIR AND WATER PERMEABILITY STUDY .........................99 4.1 Development of Air Permeameter for Asphalt Pavements.........................................99 4.2 Comparison of Field Permeameter Readings ...........................................................105 4.3 Preferential Flow Path Testing..................................................................................108
xiv
CHAPTER FIVE: CONCLUSIONS AND RECOMMENDATIONS ......................120 5.1 Summary of Findings.................................................................................................120
5.1.1 Field Study..............................................................................................120 5.1.2 Laboratory Study ....................................................................................121
5.2 Recommendations......................................................................................................123 5.2.1 Guidelines for Selection of Pavement Thickness in Wisconsin .............123 5.2.2 Recommendations on Laboratory and Field Permeability Testing Procedure ................................................................................................124 5.2.3 Recommendations for Permeability and Density Criteria for Superpave Mix Designs in Wisconsin ......................................................................127
REFERENCES……………………………………………………………………… 128 APPENDIX A……………………………………………………………………….. 132 APPENDIX B………………………………………………………………………... 135
2
TABLE OF CONTENTS
CHAPTER ONE: INTRODUCTION..............................................................................4 1.1 Background and Problem Statement.............................................................................4 1.2 Literature Review..........................................................................................................6 1.3 Research Objectives....................................................................................................23 1.4 Research Methodology ...............................................................................................24 1.5 Experimental Design...................................................................................................27 1.6 Summary .....................................................................................................................28 CHAPTER TWO: FIELD STUDY ................................................................................30
2.1 Introduction.................................................................................................................30 2.2 Equipment and Methods .............................................................................................33 2.3 Statistical Analysis of Field Studies ...........................................................................36
2.3.1 Fine Mixes ...................................................................................................41 2.3.2 Coarse Mixes ...............................................................................................53 2.3.3 Density Growth............................................................................................58
2.4 Investigation of Specification Criteria ........................................................................64 2.6 Summary of Findings from Field Study .....................................................................70 CHAPTER THREE: LABORATORY DATA ANALYSIS AND DISCUSSIONS ...73 3.1 Introduction.................................................................................................................73 3.2 Field Cores Permeability Testing................................................................................73
3.2.1 Equipment and Methods ..............................................................................73 3.2.2 Density and Permeability Results ................................................................80
3.3 Laboratory Compacted Specimen Testing..................................................................82 3.3.1 Alternatives for Laboratory Compaction Method........................................82
3.3.2 Proposed Compaction Procedure.................................................................85 3.3.3 Density and Permeability Results ................................................................88 3.4 Correlations of Lab and Field Results ........................................................................90 3.4.1 Correlation between Field Density and Lab Density ..................................90
3.4.2 Correlation between Field Permeability and Lab Permeability of Field Cores ..............................................................................................91
3.4.3 Correlation between Laboratory Permeability of Field Cores and Predicted Permeability Using Lab Compacted Specimens...................96 3.5 Summary of Findings of Laboratory Study ................................................................97
CHAPTER FOUR: AIR AND WATER PERMEABILITY STUDY .........................99 4.1 Development of Air Permeameter for Asphalt Pavements.........................................99 4.2 Comparison of Field Permeameter Readings ...........................................................105 4.3 Preferential Flow Path Testing..................................................................................108
3
CHAPTER FIVE: CONCLUSIONS AND RECOMMENDATIONS ......................120 5.1 Summary of Findings.................................................................................................120
5.1.1 Field Study..............................................................................................120 5.1.2 Laboratory Study ....................................................................................121
5.2 Recommendations......................................................................................................123 5.2.1 Guidelines for Selection of Pavement Thickness in Wisconsin .............123 5.2.2 Recommendations on Laboratory and Field Permeability Testing Procedure ................................................................................................124 5.2.3 Recommendations for Permeability and Density Criteria for Superpave Mix Designs in Wisconsin ......................................................................127
REFERENCES……………………………………………………………………… 128 APPENDIX A……………………………………………………………………….. 132 APPENDIX B………………………………………………………………………... 135
4
CHAPTER ONE
INTRODUCTION
1.1 Background and Problem Statement
It is well recognized that density that could be achieved in the field is significantly
affected by the maximum aggregate size of aggregates, the gradation, and the lift
thickness. It is also well known that permeability of asphalt mixtures is a function of
aggregate gradation, density achieved, and distribution of air voids. With the shift in
mixture designs to Superpave methods, gradations on the coarse side of the maximum
density line are being widely recommended and used. These gradations are unique in
their densification characteristics and are claimed to be more permeable. It is not clear
whether this trend is due to changes in the air voids distribution, the lower densities being
achieved, or both. This trend is of special importance to Wisconsin as the shift to
Superpave mixtures is underway. Recent studies have also shown that permeability is a
directional property such that orientation of the aggregates, which is affected by lift
thickness and level of compaction, has a significant effect on total permeability.
Wisconsin has traditionally used 75-mm dense graded HMA overlays placed in
two lifts, a 44-mm binder lift and a 31-mm surface lift. These lift thicknesses are based
on the traditional rule that lift thickness be twice the maximum aggregate size. Starting in
the year 2000, Wisconsin has decided to move from Marshall design to Superpave
5
mixture design. Superpave mixes tend to be harder to compact. Additionally, Superpave
guidelines recommend the lift thickness be a minimum of 3 times the nominal maximum
aggregate size. This move poses two problems for Wisconsin: 1) the current design
criteria for overlay thickness will result in thin-lifts of Superpave mixes that the
AASHTO Lead States Committee has reported as having problems with pavement
permeability and achieving pavement density, and 2) these mixes may be impossible to
compact in the field contributing to the permeability problem, even though they meet
laboratory density criteria.
There is a need, therefore, for a study to evaluate the potential problems and to
establish procedures to relate laboratory density to field study and to estimate or measure
permeability during mixture design. The study also needs to define the relationship
between lift thickness and aggregate gradations that will minimize the densification
problem and address the permeability concerns.
There is a previous study conducted by the University of Wisconsin-Madison
evaluating the effect of lift thickness to maximum aggregate size ratio on compaction of
Superpave mixtures (WHRP Report # 03-02). This study showed the effect of size to the
thickness ratio and indicates that density is highly dependent on size and gradation in the
laboratory when the Superpave Gyratory Compactor is used. It was also found that the
optimal size to thickness ratio varies according to the angularity and source of the
aggregates. The extrapolation of the laboratory results to the field was not achieved and
in the field the limited study could not show the same trend observed in the laboratory.
Also the project did not cover the permeability of mixtures which is an important
property that could affect pavement layers integrity and performance. Thus the field
6
validation of factors affecting density and permeability is the challenge that was
addressed in this new project, which is the subject of this report.
1.2 Literature Review
This section summarizes information on previous research related to the
permeability and density of hot-mix asphalt (HMA) mixture within the scope of the
project. The fundamental concepts of permeability are described as a background for the
permeability measurement. The equipment and test methods currently employed by
various agencies in determining the permeability both in the laboratory and in the field
study are reviewed, and the one considered the most appropriate was selected for the
research. Critical factors that need to be considered in evaluating the permeability are
discussed, and the required levels of each critical factor in the experimental design are
indicated. The study of the correlation between lab and field permeability values based
on previous studies is also summarized in this section. Additionally, the critical factors
affecting density in the field, which is considered to be the main factor influencing
permeability, are described.
1.2.1 Fundamental Concepts of Permeability
In 1856, Henry Darcy, a French civil engineer, established the fundamental
concept of permeability. His concern in the public water supply led him into the design
of permeable filter sands for water purification. He investigated the flow of water
through sand, and the parameter of his experiment was called the coefficient of
permeability or the permeability. The permeability is the rate of water flow and is
7
proportional to the hydraulic gradient. The permeability in Darcy’s law can be written
as:
ALHkAikQ ⋅
∆⋅=⋅⋅= (1.1)
where Q is the rate of flow, k is the permeability, i is the hydraulic gradient, ∆H is the
head loss across specimen, L is the length of specimen, and A is the cross-sectional area
of specimen perpendicular to direction of flow. The total head loss is the sum of
elevation head loss (He) and pressure head loss (Hp), where the pressure head is related to
the water pressure (u). Figure 1.1 shows the fundamental concept of the flow of water
through a specimen.
Figure 1.1 Fundamental Concept of Permeability Testing
8
Darcy’s law is valid for the flow through most granular materials. As long as the
flow is laminar, a linear relationship between specific discharge and hydraulic gradient is
found. Under the turbulent flow, the water flow paths are more tortuous; therefore, the
relationship becomes nonlinear. According to Darcy’s law, there are two testing methods
used to measure the permeability, a constant head method and a falling head method
(falling head or falling headwater-rising tailwater).
The system for constant head method can maintain a constant hydraulic pressure
or head to within ±5%, and the head loss across the specimen is held constant. As shown
in Figure 1.2(a), the water is allowed to flow through specimen. After an adequate
amount of water is collected over the time of test, the flow rate Q is determined. The
permeability is then calculated by:
AhQLK = (1.2)
where Q is the flow rate, L is the length of specimen, A is the cross-sectional area of
specimen, and h is the constant head shown in Figure 1.2(a).
In the falling head method, the head in the standpipe and the time are measured.
The permeability is then calculated by:
⎟⎟⎠
⎞⎜⎜⎝
⎛=
f
o
hh
lnAtaLK (1.3)
9
where a is the cross-sectional area of the stand pipe, L is the length of specimen, A is the
cross-sectional area of specimen, t is the time required for the head to fall from ho to hf, h1
is the water head at beginning of test, and h2 is the water head at end of test (Figure 1.2
(b)).
The literature review indicates that the constant head method is more appropriate
for measuring high permeable materials (K > 10-3 cm/s), and the falling head method is
more appropriate for measuring less permeable materials (K < 10-3 cm/s). The falling
head method is, therefore, a better alternative for measuring the permeability of asphalt
mixtures, for which the typical values of permeability are in the range of 10-3- 10-5 cm/s.
a) Constant Head Method
10
b) Falling Head Method
Figure 1.2 Two common methods for measuring permeability
Air permeability testing has also been utilized to quantify the permeability of dry,
porous media. Early applications were based on the falling head water permeability tests,
with a pressurized air vessel substituting for the imposed water head and the quantity of
air flow through the porous media related to the pressure drop in the air supply. To
calculate values of permeability using air flow, the form of Darcy’s law commonly used
in falling head techniques must be manipulated. The following equation provides the
intrinsic or absolute permeability (Collins 1961, Hillel 1998, Weaver 1955) of the porous
media being tested:
1
2
lna
pVLKATP p
µ ⎛ ⎞= ⎜ ⎟
⎝ ⎠ (1.4)
11
where K is the intrinsic (absolute) permeability, V is the volume of the pressure
chamber,µ is the dynamic viscosity of air, A is the cross-sectional area of sample, T is
change in time (seconds) over pressure loss, Pa is the atmospheric pressure, p1 is the air
pressure at the beginning of time measurement, and p2 is the air pressure at the end of
time measurement. The intrinsic permeability can be equated to a measurement of the
average diameter of the effective void pathways (Collins 1961, Hillel 1998). This value
of permeability is considered absolute because it is independent of the fluid flowing
through the porous medium.
The intrinsic permeability can be expressed as an equivalent hydraulic
conductivity (commonly referred to as permeability) through the following relation
(Collins 1961, Hillel 1998, Weaver 1955):
ww
w
K kg
µρ
= (1.5)
where K is the intrinsic (absolute) permeability, kww is the hydraulic conductivity
(permeability), wµ is the dynamic viscosity of water, wρ is the mass density of water,
and g is the acceleration due to gravity. Equations 1.4 and 1.5 may be combined to yield
a direct equation for calculating the hydraulic conductivity for air permeameter
measurements as:
1
2
lnww
a w
VL g pkATP pµρµ
⎛ ⎞= ⎜ ⎟
⎝ ⎠ (1.6)
12
1.2.2 Selection of Equipment and Methods of Measurements
Researchers in the past have identified several equipment and test methods to
measure the permeability of asphalt mixture both in the laboratory and in the field. The
existing equipment and their test methods are summarized in Table 1.1.
Table 1.1 Equipment and test methods used to measure the permeability of asphalt mixture
Measurement Equipment and Method
Testing Factors Researcher Comments
Falling-head type permeameter
Degree of saturation (significant effect)
Vallerga and Hicks (1968)
Apply back-pressure to ensure saturation
- Permeameter (Karol-Warner) - Falling-head method - FDOT Procedure
- Confining pressure (insignificant effect) - Testing time (insignificant effect)
Hall et al. (2000)
- Widely used for HMA - Some shortcomings found in FDOT method (no method ensure saturation)
Flexible-wall, dual mode permeameter, developed in LTRC
None
Huang et al. (1999)
- Good for determining the permeability when Darcy’s Law is not valid - Darcy’s Law is not valid for high effective porosity, and high permeability
Laboratory Permeameter
- Flexible-wall permeameter - Falling-head rising-tail method - ASTM D 5084 (Method C)
- Degree of saturation (significant effect) - Hydraulic gradient (insignificant effect) - Sidewall leakage (significant effect)
Kanitpong et al. (2001)
Apply back-pressure to ensure saturation
Field Permeameter
NCAT permeameter
None Cooley, (1998) Ease of use, repeatable, correlated to laboratory results.
13
Two-Way Permeameter
Two-Way Permeameter
Degree of permeability anisotropy of compacted or undisturbed soil
Moore, (1979)
Among these equipment and methods, the flexible-wall falling-head permeameter
developed by Florida Department of Transportation (FDOT) have been commonly
employed as a laboratory permeameter for asphalt mixtures (Mallick et al. 2001 and Hall
et al. 2000). This method, developed in 2001, is widely known as the ASTM PS 129-01
(Standard Test Method for Measuring the Permeability of Asphalt Mixtures). However,
there are some shortcomings associated with this method (Kanitpong et al. 2001). The
concern about degree of saturation is very important, because the permeability can vary
in the orders of magnitude as the degree of saturation of asphalt mix is varied. As the
degree of saturation decreases, the permeability decreases as well, since water cannot
flow through air bubbles in the voids. Since the Florida method (PS 129-01) has no
method to ensure saturation or to control the degree of saturation, the permeability
obtained from the Florida method cannot be directly applied to describe the capability of
HMA to transmit fluid. More importantly, the Florida method does not ensure a
consistent degree of saturation for HMA mixtures. That is, different degrees of saturation
can be obtained with the Florida method when testing specimens having different
characteristics. Because of the shortcomings of the Florida method, other reliable
methods were considered for this research.
Another commonly used method is the ASTM D 5084 (Standard Test Method for
Measurement of Hydraulic Conductivity of Saturated Porous Materials Using a Flexible-
Wall Permeameter). This method was selected and used by Kanitpong et al. (2001) for
asphalt mixtures because it is one of the most widely used in North America for
14
measuring the saturated permeability of porous materials having saturated permeability
less than or equal to 10-3 cm/s. In addition, the factors listed as shortcomings in the
Florida method were considered when ASTM D5084 was developed. The ASTM D
5084 method allows applying the backpressure saturation procedure that ensures the
saturation, as well as produces consistent and repeatable data, which could not be
obtained with the widely used Florida method. The ASTM D 5084 was, therefore,
selected in the laboratory study of this research study.
For the field study, the field permeameter that was designed and developed by the
National Center for Asphalt Technology (NCAT) was selected. The literature reviewed
indicated that this kind of permeameter can give a good correlation with the laboratory
permeameter, is repeatable, and is easy to use (Cooley 1998). The study from WPI
(2000) has raised some concerns about of the NCAT device; hence a modified
permeameter was developed by using NCAT device as a model. In addition to using the
modified version of the NCAT device, the research team decided to take pavement layer
cores at the same test location as the field-testing is conducted and to conduct the
laboratory permeability on the cores taken from the field for direct comparison.
1.2.3 Factors Affecting Lab and Field Permeability of Asphalt Mixtures
Several researchers identified a number of factors that can affect the permeability
of gyratory-compacted and field-compacted asphalt mix specimens. Air void content,
effective voids, gradation, NMAS, aggregate source, VMA, thickness, and void pathways
are among the mostly mentioned factors. The summary of these factors including the
range of critical values, the relationship to the permeability, and the supporting
15
researchers for laboratory studies and for field studies are summarized in Table 1.2 and
Table 1.3, respectively.
Table 1.2 Factors Affecting Permeability of Asphalt Mixtures in Laboratory Study
Variable
Name Range Of Values
Comments
Researcher
N/A Higher air void content, higher
permeability (affected by asphalt content)
McLaughlin, and Goetz (1955)
Above 5%
Good correlation with permeability
Gilbert, and Keyser (1973)
N/A - Higher air void content, higher permeability. - Size and connectivity are important. - Higher air voids, high possibility of air voids connectivity.
Abdullah et al. (1998)
At 7%, K ≈ 10-4 cm/s
Higher air void content (low density), higher permeability.
Westerman (1998)
At 4%, K ≈ 10-7 cm/s At 6-8 %, K ≈ 10-5-10-3 cm/s
- Significant Effect. - Higher air void content, higher permeability.
Kanitpong et.al (2001)
Air Voids
At 4%, K ≈ 8.5x10-7 cm/s At 8 %, K ≈ 1.2x10-4 cm/s
- Significant Effect - Higher air void content, higher permeability.
Kanitpong et.al (2002)
Dense-graded mix has lower permeability than gap-graded mix.
- No relationship b/w permeability and durability - Dense-graded mix is not the best for durability. Asphalt film thickness is more important.
McLaughlin, and Goetz (1955)
Coarser mixes, higher permeability
Coarser mixes, larger void sizes.
Abdullah et al. (1998)
Gradation
Open-graded mix, K ≈ 0.27-1.48 cm/s LA Type 508 open graded drainable base, K ≈ 2.47-3.61 cm/s Dense-graded mix, K ≈ 3x10-4-116x10-4 cm/s
Significant effect
Huang et al. (1999)
16
Finer graded, lower permeability (at constant air voids)
- Significant Effect - S-shaped gradation gives higher permeability
Kanitpong et.al. (2001)
At air voids < 8%, Fine graded mix has higher permeability than coarse graded mix at a given air void content
- Significant Effect Kanitpong et.al. (2002)
NMAS N/A No significant effect
Kanitpong et al. (2002)
Basalt > Granite > Limestone
Higher porosity of aggregate, higher permeability (at constant level of asphalt content)
Abdullah et al. (1998)
Aggregate Source
N/A Significant effect
Kanitpong et al. (2002)
N/A No relationship with permeability James (1965) N/A Higher VMA, higher permeability Abdullah et al.
(1998)
VMA
N/A No relationship with permeability Kanitpong et.al (2001)
N/A Correlated with air voids (limited data)
Kanitpong et.al (2001)
N/A Decrease in permeability with increase in thickness
Mallick et al. (2001)
Thickness
N/A No significant effect (From statistical view point)
Kanitpong et al. (2002)
Void Pathways Higher connectivity, higher permeability (at constant air voids content)
- Not straight and vertical, but convoluted towards to the perimeter of specimens. - Field cores have higher interconnectivity than SGC samples.
Hall et al. (2001)
17
Table 1.3 Factors Affecting Permeability of Asphalt Mixtures in Field Study
Variable Name
Range Of Values
Comments
Researcher
At 10%, K ≈ 150 ml/min (2 cm/s)
- Permeability not exceed 150 ml/min will be low enough to prevent access moisture.
Zube (1962) (Field test)
At 7%, K ≈ 10-3 cm/s (for coarse-graded)
- Permeability limit not more than 10-3 cm/s is suggested in the in-place Superpave mix pavement permeability - Air void structures in gyratory sample, and field compacted core are not comparable (at same air voids level)
Choubane et al. (1998) (Laboratory test of field cored sample)
At 7%, K increased significantly
Significant effect
Mallick et al. (2001)
Air Voids
The critical values of air voids depend on NMAS 7.7% for 9.5 mm NMAS 7.7% for 12.5 mm NMAS 5.5% for 19.0 mm NMAS 4.4% for 25.0 mm NMAS (for coarse-graded Superpave mix)
Significant effect Cooley et al. (2001)
Coarse-graded has higher interconnectivity of voids.
- In-place air voids of coarse-graded appear to have greater interconnectivity than fine-graded (at same air voids level)
Choubane et al. (1998) (Laboratory test of field cored sample)
Gradation
No difference occurred between coarse and fine graded mixes
- Can not compare because higher air voids in fine graded mix, and different in thickness
Mallick et al. (2001)
NMAS At air voids = 6%, and for coarse-graded, 9.5 mm NMAS, K ≈ 6x10-5 cm/s 12.5 mm NMAS, K ≈ 40x10-5 cm/s 19.0 mm NMAS, K ≈ 140x10-5 cm/s 25.0 mm NMAS, K ≈ 1200x10-5 cm/s
- Significant Effect - At given air void content, permeability increased by one order of magnitude as the NMAS increased.
Mallick et al. (2001)
18
For coarse-graded Superpave mix, 9.5 mm NMAS, K ≈ 100x10-5 cm/s 12.5 mm NMAS, K ≈ 100x10-5 cm/s 19.0 mm NMAS, K ≈ 120x10-5 cm/s 25.0 mm NMAS, K ≈ 150x10-5 cm/s
Significant effect Cooley et al. (2001)
Aggregate Source
None None None
VMA None
None None
4 times NMAS (required for coarse-graded Superpave mixes) Adequate density results in adequately low permeability
- Because criteria for fine-graded Marshall mixes may not be adequate for coarse-graded Superpave mixes. Westerman (1998)
Choubane et al. (1998) (Laboratory test of field cored sample)
Thickness Min lift thickness ≥ 51 mm, or 4 times NMAS
Significant effect Cooley (2001)
Based on the summary, air void content was found to be the most critical factor
that can affect the permeability both in the laboratory and in the field study. As the air
voids increase, the permeability also increase (McLaughlin and Goetz 1955, Zube 1962,
Westerman 1998, Choubane et al. 1998, Gilbert and Keyser 1998, Abdullah et al. 1998,
Kanitpong et al. 2001, Mallick et al. 2001). However, a recent detailed study indicates
that in measuring the permeability of asphalt mixtures, the total volume of voids is not as
important as the connectivity of voids (Huang et al. 1999). Therefore, the relationship
between the effective voids content, which is the ratio of voids to be drained under
gravity to the total volume of mixture, and the permeability, was also evaluated by some
of the researchers (Huang et al. 1999, Cooley et al. 2002, Al-Omari et al. 2002).
Gradation also plays as a significant role in the permeability. Coarse graded mix
contains larger void sizes, and has higher possibility for connectivity of voids, hence
19
resulting in higher permeability (McLaughlin and Goetz 1955, Choubane et al. 1998,
Huang et al. 1999, Abdullah et al. 1998). S-shaped gradation was also found to have
higher permeability compared with other coarse graded mixes (Kanitpong et al. 2001).
While the NMAS was not found to have a significant effect on the permeability of
laboratory compacted specimens (Kanitpong et al. 2001), it significantly affects the field
permeability (Mallick et al. 2001). This result could point out the problem of the
discrepancies between permeability of laboratory compacted specimens and the field
specimens.
Aggregate source is shown to have a significant effect on the permeability
(Abdullah et al. 1998, and Kanitpong et al. 2002). Aggregate shape affects size of voids,
shape, and connectivity of voids, and hence, directly influences the permeability.
However, there exist the inconsistent results regarding to the effect of percent VMA on
the permeability in the literature.
Lift thickness is also a questionable factor. Some researchers stated that the lift
thickness significantly affects in the field density and permeability (Choubane et al. 1998,
Mallick et al. 2001). Unfortunately, this finding could not be observed as a significant
factor for laboratory compacted specimens (Kanitpong et al. 2002). It can only be
concluded that further study is necessary to investigate the effect of lift thickness, and the
correlation between the laboratory and field specimen need to be evaluated.
Void pathway was indicated as an important variable that need to be addressed.
Hall (2000) found that most of void pathways are not straight and vertical, but convolute
towards the perimeter of specimens. In addition, he also found that the field-cored
specimens have higher connectivity of voids than the gyratory compacted specimens.
20
1.2.4 Relationship Between Lab and Field Measurements
This section includes the discussion from previous study that evaluating the
relationship between lab and field permeability, and the relationship between the
permeability of field cores and gyratory compacted specimens.
Relationship Between Lab and Field Permeability Measurements
Cooley’s research (2002) includes the comparison study between the lab and field
permeability measurement (Cooley et al. 2002). The laboratory permeability tests were
conducted on cores cut from the pavement sections for which the in-place field
permeability was measured. They found that the relationship between field and
laboratory permeability results is not simple. At permeability less than 500x10-5 cm/sec,
the lab permeability is higher than the field permeability. However, they indicated that
this result was not as expected, since the field results should provide higher permeability
because water can flow from the field permeameter in any direction, while laboratory
permeameter restricts water to flow in only one direction. The field test was, therefore,
expected to obtain higher permeability. A possible explanation for this result is that, at
permeability above 500x10-5 cm/sec, asphalt mixes have a high percentage of
interconnected air voids. In the field, these interconnected air voids may or may not be of
a length that they allow water to flow. On the other hand, the laboratory permeameter
may allow a single large interconnected air void that extends within the asphalt specimen
and result in high laboratory permeability.
According to their conclusion on the relationship between lab and field
permeability measurements, it is indicated that both methods provide similar results at
permeability values that are not excessive. Cooley et al. (2001) suggested that field
21
permeability values should be less than 150x10-5 cm/sec. Their study suggests that the
field permeameter provide reasonable results and are comparable to the controlled lab
permeability test method. The advantages of field permeameter are that it provides more
rapid test results and is nondestructive.
Relationship Between Permeability of Field Cores and Gyratory Compacted Samples
Because of the differences in air void distribution of the laboratory and field
compacted samples, similar interconnected void structures are unlikely. Cooley et al.
(2002) conducted a study to evaluate the relationship between permeability and density
with lab and field compacted mixes. Two techniques were used in their study: laboratory
permeability measurements on samples compacted using the gyratory compactor and
water absorption determined with AASHTO- T166 and the Corelok device.
In Cooley’s study, the Superpave Gyratory Compacted (SGC) samples could not
be produced at the exact same air void levels as the field cores, therefore, the relationship
between air voids and lab permeability was determined for each of the three NMAS. The
9.5 mm mix indicated that there is a strong relationship between air voids and
permeability. However, the relationships between permeability and density are different
between two specimen types (lab-compacted vs field compacted). The results of field
specimens show higher permeability at a given air void content than the lab specimens.
For the 12.5 mm and 19 mm NMAS mix, there is a good relationship between density
and lab permeability for both the field cores and the SGC specimens. The limited data in
their study indicated that SGC samples could be used to estimate the field compaction
level required to produce an impermeable mix.
22
In addition, the relationship between water absorption and permeability was
evaluated to identify a parameter that would indicate potential permeability problems in
the field. It seems intuitive that the percentage of water permeable voids should be
related to the available flow paths for the water and in turn to permeability. The results
from this study showed a reasonable relationship between water absorption during
AASHTO T-166 and water permeable voids from Corelok testing and permeability
results (both field and lab). This may be used as a quick screening test to identify
pavements that may be permeable.
1.2.5 Factors Affecting Density of HMA During Construction
Several studies have been reviewed for identifying the important factors that
could affect the reaching of required density of HMA during the construction. Some of
these factors will be considered as independent variables in the experimental design for
this study, and they are summarized as shown in Table 1.4.
Table 1.4 Summary of factors affecting density of HMA during construction
Variable
Type Comment
Gradation Fine Coarse
Quantitative measure need to be defined between fine and coarse materials used in the mixture
Aggregate Angularity Crushed aggregate (angular particles) Natural aggregate (round particles
Quantitative measure, i.e., % crushed faces and % crushed particles needed to be determined
Thickness Thickness to NMAS Ratio Study by Paye 2001 Compaction Force N/A Resultant and pressure
applied to the mat are measured with simple statics and geometry
23
Roller Types Vibratory Roller Pneumatic Tire Roller Static Steel Roller (Cold Roller)
Specific contribution of certain roller types to densification under varying mat thicknesses is important (Paye 2001)
Base Type Concrete Milled Asphalt CABC Rubblized Concrete
Found significant in (Paye 2001)
Temperature N/A Decreasing temperature increases the resistance of the asphalt mix to densification
1.2.6 Summary of Literature Review
The review of literature on permeability measurement of HMA both in the
laboratory and in the field has resulted in the following action items:
• Selecting the flexible-wall permeameter using the ASTM D5084 method,
and the NCAT permeameter for measuring laboratory and field
permeability, respectively, in this study.
• A number of variables that could affect the density and permeability of
HMA are considered and included in the experimental design.
• The findings of some literature that included the permeability and density
criteria of HMA will be compared to the final results of the study.
1.3 Research Objectives
The objectives of this research are as follows:
1. Determine the influence of maximum aggregate size, lift thickness, and
aggregate source on the density and permeability of asphalt mixtures designed
according to the Superpave criteria.
24
2. Develop guidelines for the selection of pavement layer thickness based on
nominal maximum aggregate size and gradation for use in Wisconsin.
3. Evaluate the effect of void characteristics, arrangement, and interconnectivity
on permeability.
4. Recommend laboratory and field permeability testing procedures and
equipment for design and quality control of Superpave mixtures in Wisconsin.
5. Recommend permeability and density criteria for Superpave mixture designs
in Wisconsin based on traffic, lift thickness, field drainage and moisture
conditions.
1.4 Research Methodology
The research methodology used is illustrated in Figure 1.3. The research plan is
divided into seven major tasks, which are described as follow:
Task 1: Literature Review on Density and Permeability of Superpave Mixes
A literature review was conducted to document published information and results
of studies conducted at the national and regional level as related to this project. The
result of this task was summarized in section 1.2 in this chapter. According to the
literature review study, the most appropriate equipment and methods for measuring
density and permeability were selected for this project. The critical factors that need to
be covered in the laboratory and in the field study were also identified and the required
levels of each critical factor in the experimental design were selected as discussed in the
next section.
25
Task 2: Identify critical variables and select commercial HMA plants with
consistent aggregate sources
In this task, the critical variables that affect the density and permeability of HMA
were initially defined and used in the experimental design. The research team, in
collaboration with the Wisconsin DOT and the representative of the asphalt paving
industry, then selected HMA plants with consistent aggregate sources. The major
aggregate sources representing the most widely used aggregates in Wisconsin pavements
were selected. Other critical variables such as gradation and nominal maximum aggregate size
were also considered in the selection of Superpave mix and materials used in the study.
Task 3: Identify projects for field comparisons
In this task a set of projects were selected that allow measuring the effect of
different variables identified in the experimental results. These projects include
Superpave mixtures with different nominal maximum aggregate size, gradations,
aggregate sources, lift thickness and sub-surface layers. The selection was based on a
review of WisDOT projects and other projects that the asphalt industry is involved in.
The characteristics of each project were documented first, and based on specific criteria;
the projects were ranked and matched with required factors to be studied. The highest
ranked projects were selected and reviewed with the members of the flexible pavement
TOC to finalize the list and contact the contractors involved.
Task 4: Conduct Field and Laboratory Studies
In the field study, the in-place densities were measured by using nuclear gauges,
and the field permeability was performed immediately after the density was measured.
The field permeability was measured by using a falling-head permeameter similar to
26
NCAT device. The field cores were then taken to laboratory after the permeability was
completed. The loose mix from each project was taken to the lab for producing the
laboratory compacted specimens. In the laboratory study, the Superpave gyratory
compactor (SGC) was used to compact the specimens from loose mixes at the same
density as the field cores. The lab permeability was then measured for field cores and
lab-compacted specimens. The relationships between field permeability, lab permeability
of field cores, and lab permeability of lab-compacted specimens were determined from
the results obtained in this task.
Task 5: Air and Water Permeability Studies
This task includes the field permeability tests conducted on newly constructed
asphalt pavements using the NCAT water permeameter and the ROMUS air
permeameter. The ROMUS air permeameter was designed and constructed by Jay
Schabelski during this study to provide a more efficient alternate to the NCAT device and
to be suitable for field testing of asphalt pavement types investigated during this study.
This device was furnished to the project team and a comparison of field permeability
results obtained with both devices was established in this task. It is believed that the
ROMUS air permeameter may be better suited to in-place permeability testing of asphalt
pavements as the device produced more efficient and repeatable measures than the NCAT
water permeameter for all pavement types investigated.
Task 6: Analyze Data and Prepare Guidelines
According to the results in Task 4 and 5, the data was analyzed and the guidelines
were prepared. Statistical analysis was used to establish the relationship between
permeability, density and the controlled variables. These variables included lift
27
thickness, nominal maximum aggregate size, gradation, aggregate source, sub-surface
layers, and other factors that might be found through the research. A relationship
between permeability, density and lift thickness to aggregate size was evaluated.
Task 7: Prepare and Submit Final Report
This final report was written to include work conducted in Tasks 1 to 6 of this
resercah study. It also includes the guidelines for how to select the effective pavement
thickness corresponding to the permeability and density criteria. The primary product of
this research is a table describing the relationship of recommended Superpave pavement
thickness, nominal maximum aggregate size and gradation to the permeability and
density of Superpave mixes. The second product is a recommendation for the laboratory
testing procedures to predict the permeability in the field. These products are included in
a final report that reflects the basis for recommended guidelines and that documents the
research effort.
1.5 Experimental Design
To accomplish the research objectives, the experimental design selected included
the following experimental variables:
Response Variables
• Density
• Permeability
Controlled Variables
• Sub-surface layers: Strong Base (Concrete) and Weak Base (HMA and CABC)
28
• Aggregate sources: Limestone and Granite
• Gradation: Coarse and Fine
• Nominal maximum aggregate size (NMAS): 9.5 mm, 12.5 mm, 19 mm, 25 mm
• Lift Thickness to NMAS ratio: In range of 3-5
1.6 Summary
This report is organized into five chapters. Chapter 1 includes the background,
problem statement, literature review, objectives, research methodology, and research
scope. Chapter 2 includes the field data analysis and discussions. The results from the
field study are described in details and the effect of different variables on the field density
and permeability is determined. The guidelines for the selection of pavement layer
thickness based on nominal maximum aggregate size and gradation are also developed in
this chapter. Chapter 3 includes the results of the laboratory study. The relationship
among the field permeability, lab permeability of filed specimens, and lab permeability of
lab-compacted specimens is evaluated. The laboratory testing procedure for predicting
permeability in the field is also recommended in this chapter. Chapter 4 contains the
analysis and comparison of the air and water permeability results. Chapter 5 includes a
summary of findings, the conclusions from this study, and the recommendations for
future research.
29
Figure 1.3 Research Methodology
Laboratory Study
Obtain loose mix from field
Compact samples with Superpave gyratory compactor at:
• Ndes • In the range of field density • Varying thickness • Varying amount of material used
Perform laboratory permeability testing
Select sections, measure in-place densities by using nuclear gauges, perform permeability testing by using modified NCAT device and ROMUS device and collect cores at the same spot of testing (at least 6 cores per project) to test in laboratory
Bring cores and measure density, permeability, and void pathways in laboratory
Collect all data and analyze
Field Study
Conduct Laboratory and Field Study
Work with WisDOT to find projects that correspond to experimental design
Experimental Design
Identify Variables Response Variables
• Density • Permeability
Independent Variables • Sub-surface layers • Aggregate sources • Gradation • NMAS • Lift Thickness
30
CHAPTER TWO
FIELD STUDY
2.1 Introduction
The field portion of the study was designed to determine factors that influence
density and permeability of WisDOT asphalt pavements designed according to Superpave
criteria. From the literature review, variables thought to have an effect on density and
permeability were selected. Three primary variables evaluated between projects were:
(1) Gradation (coarse and fine), (2) Source (gravel and limestone), and (3) Base Stiffness
(rigid and flexible). A total of eight project combinations allowed a direct evaluation of
these variables: [2 gradations x 2 sources x 2 base types] = 8 projects. Ndes was an
additional between-project variable, but was not directly controlled during project
selection. Six variables evaluated within each project, included: (1) Nominal Maximum
Aggregate Size (NMAS), (2) layer thickness, (3) layer thickness-to-NMAS ratio
(t/NMAS), (4) density, (5) fine-graded mix aggregate ratios, and (6) roller set-up.
Gradation was classified coarse as follows:
25.0-mm NMAS: Less than or equal to 25% passing the 2.36-mm sieve
19.0-mm NMAS: Less than or equal to 30% passing the 2.36-mm sieve
12.5-mm NMAS: Less than or equal to 35% passing the 2.36-mm sieve
9.5-mm NMAS: Less than or equal to 40% passing the 2.36-mm sieve
Although there are several types of aggregate sources used in Wisconsin highway
construction, only gravel and limestone were selected because of their widespread use
and limited resources available for the study. The field study balanced statistical
requirements with resource availability, and it was concluded that it was more beneficial
31
to have fewer sources and a greater amount of testing within each source. When
comparing the physical shape of the two aggregates, gravel has a more round, cubical
appearance, while limestone has a sheared-plane, multi-faced appearance.
Base type was classified rigid if it was Portland Cement Concrete (PCC), and
flexible if it was asphalt pavement (milled or existing) or crushed aggregate. No testing
was performed on rubblized PCC bases.
Table 2.1 provides a matrix of projects with each specific variable combination.
It was not possible to collect data from all variable combinations, in particular projects
having a gravel source and rigid base. Two gravel/rigid candidate projects were
warranted pavement and WisDOT staff allowed no coring, thus precluding those projects
from the study. Findings from the 2002 data allowed the research to screen the most
significant variables, and allow more detailed experimentation during the 2003 data
collection phase. An analysis of field data collected from the 2002 paving season found a
wide amount of variation in field data, and doubling the combinations allowed greater
resolution in the data. For several variable combinations, two projects were used for data
collection to strengthen the data set and analysis.
32
Table 2.1 Project Matrix
Project Gradation Source Base NMAS, mm Ndes (1) (2) (3) (4) (5) (6)
Wis. Ave. Fine Limestone Rigid 19, 12.5 75 I-894 19, 12.5 100
USH-10 Fine Limestone Flexible 19, 12.5 60 STH-21 19, 12.5 100
--- Fine Gravel Rigid --- --- --- --- ---
STH-23 Fine Gravel Flexible 19, 12.5 75 USH-8 19, 12.5 75
I-43 Coarse Limestone Rigid 19 100 USH-20 (ILL.) 19, 9.5 70
I-94 Coarse Limestone Flexible 25 125 --- --- --- --- Coarse Gravel Rigid --- --- --- --- ---
STH-17 Coarse Gravel Flexible 25 75 --- --- ---
Within each project, variation in thickness and density were produced from
natural field construction variation or fabricating the variation at locations within the
project by adjusting construction operations. NMAS was varied by testing different
layers within a project. However, on both the I-94 and STH-17 coarse-graded projects, a
fine-graded surface mix was paved and no testing was performed on this layer. Roller
operations were documented, including such factors as roller type (steel drum or
pneumatic tire), vibratory compaction (yes or no), mat temperature, and number of
passes.
33
2.2 Equipment and Methods
Field data collection for each project involved five primary steps: (1) coordination
and test site identification, (2) density growth testing, (3) NCAT permeameter testing, (4)
air permeameter testing, and (5) pavement coring. Field data collection occurred the day
of paving, or a few days after paving before traffic was permitted on the test sites. All
testing was conducted before rainfall.
2.2.1 Coordination and Test Site Identification
Projects were selected and coordinated with WisDOT and contractor personnel.
Effort was made to conduct testing with minimal disruption to scheduled construction
activities. Testing time on a project normally required a minimum of four hours per
layer. Contractor mix designs and loose mix were collected for the laboratory component
of this study (see laboratory section of report).
Six test sites were selected for each NMAS layer within a project. For example,
STH 23 had six test sites in the 19-mm NMAS bottom layer, and six test sites in the 12.5-
mm NMAS top layer. A minimum of six test sites was chosen per layer to optimize the
number of data points within the time and resource constraints of field testing. On
several projects, more than six test sites per layer were field tested to guard against core
damage, ensure thickness and density variation, or other factors.
Two variables controlling test site selection were layer thickness and density. The
paving crew provided an estimate of planned layer thickness for the day’s paving, and
sites producing the widest range and median were selected. Two test sites each were then
chosen from the minimum, median, and maximum thickness areas. Test sites with
34
t/NMAS ratios below 3 and above 5 (outside the range in Subsection 405.3.9.2 of the
Specifications) were given priority in an effort to broaden the data range and help
understand this effect on density and permeability. In many cases, initial test sites were
discarded due to insufficient density (generally below 90%), surface segregation, or an
uneven surface profile from roller wheels. Data on the USH-41/Lannon Road
Intersection project was discarded due to median density values below 90%, segregation,
delayed paving schedule, and other project factors. Compacted layer thickness was
estimated from the loose layer thickness. Maximum surface slope for all test sites was
limited to 4%.
2.2.2 Density Growth Testing
Density growth testing was conducted to measure the compactability of the
pavement layer from typical project variables, such as NMAS, layer thickness, roller
type, change in density from screed to finish roller, mat temperature, and vibratory
application. Multiple 15-second readings were taken with the nuclear density gauge at
each test site behind the paver screed and after series of roller passes. Vibratory setting
(on or off) and pavement temperature were recorded after the roller pass. In several
cases, it was not possible to collect data after every pass due to safety concerns.
2.2.3 NCAT Permeameter Testing
After density growth testing, the pavement was allowed to cool naturally for
permeability testing with the NCAT permeameter device. The pavement was generally
tested below a surface temperature of 125°F to ensure an adequate seal. The
35
permeameter was centered within the rectangular base used for nuclear density growth
testing, sealant was applied between the pavement and permeameter base, a 20-kg weight
was added to prevent uplift force from the water head, then the pavement was saturated.
Several trials were conducted at each test site for repeatability information and to
incorporate testing variability into the analysis.
2.2.4 Air Permeability Testing
Air permeability testing was conducted using the ROMUS device at locations
selected for NCAT water permeability testing and pavement coring. Air permeability
testing was conducted immediately preceding water permeability testing to eliminate the
potential for water infiltration into the air permeameter. Test locations were displaced
approximately 6 inches longitudinally from pre-selected water permeability test locations
to minimize the potential for the grease seal produced by the ROMUS device to
contaminate the surface to be tested with the NCAT device.
2.2.5 Pavement Coring
Upon completion of air and water permeameter testing, cores were cut in the
exact location of the water permeameter test. The six-inch diameter circular seal residue
from the NCAT permeameter served as a guide for positioning the core drill. After the
core was cut and removed from the pavement, it was marked and transported to the lab
for bulk density testing using the Corelok device. If a core was damaged, a substitute test
site was used to ensure a minimum of six test sites per layer.
36
2.3 Statistical Analysis of Field Studies
A formal analysis of variance (ANOVA) was conducted to measure sources of
variation influencing density and water permeability. The F-test was used to determine
statistical significance of each variable and reported in three ranges (<0.01; 0.01 to <0.05;
and 0.05 to 0.10) to help understand the degrees of significance, rather than significance
at an absolute level, such as 0.05. In some cases, it was not possible to test a variable due
to lack of data, or a high degree of collinearity between variables that would have not
made it possible to discern between significant variables. Main effects and two-way
interactions were tested, and three-way interactions and higher were not tested to
conserve degrees of freedom for significance testing (six test sites per project limits total
pooled observations and significance sensitivity).
Tables 2.2 and 2.3 provide a summary of statistical significance tests of variables
for field permeability and final pavement density, respectively. Because of significant
differences in permeability between fine-graded and coarse-graded mixes, each was
analyzed separately. For each gradation, all project data were pooled to test the variable
of interest. Appendix A provides results of significance tests for individual projects. It
must be noted that these findings are strictly limited to data collected in this study, and
may not represent all mixes constructed using WisDOT design standards.
Permeameter test variability was relatively high for fine-graded mixes. Test
variability of the NCAT permeameter, as a percentage of total individual project
variability, ranged from 5% to 87% on fine-graded mixes, and from 3% to 13% on
coarse-graded mixes. These measures indicate testing of coarse mixes is more repeatable
than fine mixes.
37
Table 2.2 Statistical Significance Results for Field Permeability
Variable Permeability Permeability Fine Mix Coarse Mix
(1) (2) (3) Main Effects
Base 2 levels (rigid, flexible) *** *Base 3 levels (PCC, HMA, CABC) *** *Source *** no testDensity *** *Ndes *** no testNMAS N/S no testThickness *** **Thickness/NMAS Ratio *** **
Aggregate Ratios 1. Ratio, Passing No.4 CA/FA * - - - 2. Fine Aggregate Angularity N/S - - - 3. Ratio, (P1/2 - P3/8)/(P4 - P8) *** - - - 4a. Bailey 1 N/S - - - 4b. Bailey 2 N/S - - - 4c. Bailey 3 ** - - -
Interactions - Significant Only Base x Source ** no testBase x Ndes *** no test Thickness x Base *** no testThickness/NMAS x Base *** N/SThickness x Source *** no testThickness/NMAS x Source *** no testThickness x Ndes * ***Thickness/NMAS x Ndes ** ***Thickness/NMAS x Ratio 1 * - - - Thickness/NMAS x Ratio 3 *** - - - Thickness/NMAS x Ratio 4c ** - - -
Density x Base *** *Density x Source *** no testDensity x Thickness * **Density x Thickness/NMAS ** **Density x Ndes *** ***Density x Ratio 1 ** - - - Density x Ratio 2 *** - - - Density x Ratio 4a *** - - -
38
Density x Ratio 4b *** - - - Density x Ratio 4c *** - - -
Significance Levels: N/S = Not Significant; * = 0.05 < p-value < 0.10;** = 0.01 < p-value< 0.05; *** = p-value < 0.01 no test = variable had collinearity with other variable(s); - - - = variable not tested
39
Table 2.3 Statistical Significance Results for Field Density
Variable Final Density Final Density Fine Mix Coarse Mix
(1) (2) (3) Main Effects
Base 2 levels (rigid, flexible) N/S ***Base 3 levels (PCC, HMA, CABC) - - - - - - Source *** ***Ndes *** no testNMAS *** no testThickness ** N/SThickness/NMAS Ratio N/S N/SPassing 4.75mm *** no testPassing 75um *** no testLab Voids *** no testVMA *** no testVFA N/S no testAC% ** no test
Interactions – Base, Thickness, t/NMAS only
Base x Source no test no testBase x Ndes no test no test Thickness x Base *** ***Thickness/NMAS x Base *** N/SThickness x Source N/S ***Thickness/NMAS x Source N/S N/SThickness x Ndes ** no testThickness/NMAS x Ndes *** no testThickness x P475mm *** no testThickness/NMAS x P475mm *** no testThickness x P75um N/S no testThickness/NMAS x P75um N/S no testThickness x Voids * no testThickness/NMAS x Voids no test no testThickness x VMA N/S no testThickness/NMAS x VMA no test no testThickness x VFA ** no testThickness/NMAS x VFA no test no testThickness x AC% N/S no testThickness/NMAS x AC% N/S no test
40
Significance Levels: N/S = Not Significant; * = 0.05 < p-value < 0.10;** = 0.01 < p-value< 0.05; *** = p-value < 0.01 no test = variable had collinearity with other variable(s); - - - = variable not tested
41
The following sections are graphical presentations and interpretations for fine-graded and
coarse-graded mixes, respectively, to support findings from the statistical analysis.
2.3.1 Fine Mixes
A. Base. Two tests were conducted for base type: a 2-level test for rigid and flexible,
and a 3-level test for PCC, HMA, and CABC. In both cases, base and Ndes had
an effect on permeability. Figure 2.1 provides the relationship between
permeability and the three base types. Source and Ndes data were broken down
to show their relationship with base. The ‘Base*Ndes’ interaction was
significant, and this is readily shown with higher permeability Ndes=100 data
points on rigid bases. This infers that high Ndes mixes may be more difficult to
compact on rigid bases, thus causing a more permeable pavement.
Figure 2.1 Field Permeability and Base Type (Fine Mixes)
Fine Gradation, Base and Ndes
0500
10001500200025003000
0 1 2 3 4
Base Type (1=PCC, 2=HMA, 3=CABC)
Fiel
d Pe
rm.,
kx10
-5
cm/s
ec
Gravel,Ndes=75Limestone,Ndes=60Limestone,Ndes=75Limestone,Ndes=100
42
B. Source. Limestone, when compared to gravel, had a greater influence on changes
in permeability, as shown in Figure 2.2. When compared across a similar density
range, say 90% to 95%, limestone-source pavements were more permeable.
Several two-way interactions also measured the influence that aggregate source
had on permeability. The ‘Density x Source’ interaction suggests that limestone
was more difficult to compact, thus producing lower density and higher
permeability. Likewise, the ‘Thickness x Source’ interaction and ‘Base x Source’
interaction also indicated that limestone was more sensitive to thickness and base
type when trying to achieve density, thus creating a response in permeability.
Fine Gradation, Source and Density
0500
10001500200025003000
87 89 91 93 95 97
Core Density, %
Fiel
d Pe
rm.,
kx10
-5
cm/s
ec
GravelNdes=75LimestoneNdes=60LimestoneNdes=75LimestoneNdes=100
Figure 2.2 Field Permeability and Source (Fine Mixes)
C. Density. Lower density pavements were more permeable for limestone-source
mixes, while no trend was observed for gravel-source mixes, as illustrated in
43
Figure 2.2. The ability to achieve final density for fine-graded mixes was
influences by source, Ndes, NMAS, thickness, passing 4.75mm and 75um sieves,
lab voids, VMA, and %AC. The layer t/NNMAS ratio was not statistically
significant, while the interactions of layer thickness with base, Ndes, passing
4.75mm sieve, and voids were all found to be statistically significant.
D. Ndes. Higher Ndes mixes were more permeable, as shown in Figure 2.3. The
‘Density x Ndes’ interaction also indicates that higher Ndes mixes were more
difficult to compact, producing a higher permeability rate. To further support
this, statistical results confirm Ndes had a significant affect on achieving final
density.
Fine Gradation, Ndes and Source
0
500
1000
1500
2000
2500
3000
50 60 70 80 90 100 110
Ndes
Fiel
d Pe
rm.,
k x
10-5
cm
/sec
GravelLimestone
Figure 2.3 Field Permeability and Ndes (Fine Mixes)
44
E. Thickness. For limestone-source mixes, layer thickness in the range of 2 to 3
inches was more permeable (see Figure 2.4). Gravel-source mixes had little effect
across all thickness ranges. Research findings in the report found that lab
permeability of field cores was higher for thin layers, and lower for thick layers.
The difference between field and lab tests may be the confinement provided at the
bottom of the field layer.
Fine Gradation, Thickness
0
500
1000
1500
2000
2500
3000
1 2 3 4
Layer Thickness, inches
Fiel
d Pe
rm.,
kx10
-5 c
m/s
ec GravelNdes=75
LimestoneNdes=60
LimestoneNdes=75
LimestoneNdes=100
Figure 2.4 Field Permeability Thickness (Fine Mixes)
Figure 2.5 illustrates the relationship between thickness and density for both
sources, where Ndes=100 limestone mixes had lower final density than gravel
mixes and lower Ndes limestone mixes across a range of layer thickness. A slight
45
trend can be observed between thickness and density where thinner layers had a
lower density, and thicker layers had a higher density.
Fine Gradation, Thickness vs. Density
86889092
949698
0 1 2 3 4 5
Layer Thickness, inches
Cor
e D
ensi
ty, %
GravelNdes=75
LimestoneNdes=60
LimestoneNdes=75
LimestoneNdes=100
Figure 2.5 Layer Thickness and Mat Density (Fine Mixes)
F. Thickness/NMAS Ratio. Higher t/NMAS ratios produce increased permeability
for limestone mixes with Ndes = 100 (See Figure 2.6). No trends were observed
for the other limestone mixes (Ndes = 60, 75), nor for the gravel mixes. T/NMAS
ratios below 2 for both sources produced lower final density. In addition, lower
density values for limestone mixes were observed below a t/NMAS ratio of 3, as
shown in Figure 2.7. These results suggest that ratios less than 2 have the ability
to impact final pavement density.
46
Fine Gradation, Thickness/NMAS Ratio
0
500
1000
1500
2000
2500
3000
0 1 2 3 4 5 6
Thickness/NMAS Ratio
Fiel
d Pe
rm.,
kx10
-5
cm/s
ecGravelNdes=75
LimestoneNdes=60
LimestoneNdes=75
LimestoneNdes=100
Figure 2.6 Field Permeability and Layer Thickness/NMAS Ratio (Fine Mixes)
Fine Gradation, Thickness/NMAS vs. Density
86889092949698
0 1 2 3 4 5 6
Thickness/NMAS Ratio
Cor
e D
ensi
ty, % Gravel
Ndes=75LimestoneNdes=60LimestoneNdes=75LimestoneNdes=100
Figure 2.7 Layer Thickness/NMAS Ratio and Mat Density (Fine Mixes)
G. NMAS. Figure 2.8 shows scatter among 12.5-mm and 19-mm NMAS mixes, and
no trend or difference in mean level was observed. The statistical analysis and
47
this figure are in disagreement with the NCHRP 9-27 study, and other studies in
Florida, Virginia, and Maine, that found NMAS to be a significant factor affecting
permeability (Cooley et. al 2001).
Fine Gradation, NMAS
0
500
1000
1500
2000
2500
86 88 90 92 94 96 98
Core Density, %
Fiel
d Pe
rm.,
kx10
-5 c
m/s
ec
12.5-mm NMAS19-mm NMAS
Figure 2.8 Field Permeability and NMAS (Fine Mixes)
H. Aggregate Ratios. Four aggregate ratios were evaluated.
(1) Ratio of Percentage Passing the No. 4 sieve (P4 Ratio) contributed by
coarse aggregates and sand. For this report, the P4 Ratio is defined as the
P4 fraction contributed by the sands divided by the P4 contributed by the
coarse aggregate. For example, an aggregate blend containing 40% coarse
aggregate (P4=17%) and 60% sand (P4=95%) would have a total
P4=63.8%. Of these P4 materials, 10.6% were contributed by the coarse
aggregate and 89.4% from the sands, yielding a P4 Ratio of 8.4.
Statistical analysis found moderate significance in mean levels among
ratios; however, the plot in Figure 2.9 shows no specific trend.
48
Fine Gradation, Ratio %PNo.4 FA/CA
0500
10001500
20002500
3000
0 10 20 30 40 50
Ratio %Passing No. 4, FA portion / CA portion
Fiel
d Pe
rm.,
k x
10-5
cm/s
ec GravelLimestone
Figure 2.9 Field Permeability and Ratio of Percentage Passing No. 4 Sieve, Fine Aggregate Portion / Coarse Aggregate Portion (Fine Mixes)
(2) Fine Aggregate Angularity. FAA did not have an effect on permeability.
FAA test results among projects ranged from 41.4 to 46.8.
(3) Ratio of (%P1/2 - %P3/8) / (%PNo.4-%PNo.8). Figure 2.10 shows a
relationship for Limestone, but no trend for Gravel. For a fine crushed-
limestone mix, the relative contribution of a narrower gap between Percent
Passing ½” sieve (%P1/2”) and Percent Passing 3/8” sieve (%P3/8”), or
wider gap between Percent Passing No. 4 sieve (%P4) and Percent Passing
No. 8 sieve (%P8), produced a more permeable mat. Figures 2.11 and
2.12 illustrate higher permeability as the gaps increase between coarse
aggregates (%P1/2” and %P3/8”) and fine aggregates (%P4 and %P8),
49
respectively. This suggests relative differences in these sieves may have
an effect on internal void structure, and measured permeability, of the
compacted material.
Fine Gradation, Ratio of Coarse Sieve Delta / Fine Sieve Delta
0.0
500.0
1000.0
1500.0
2000.0
2500.0
3000.0
0.00 0.20 0.40 0.60 0.80 1.00 1.20
Ratio (%P1/2 - %P3/8) / (%P4 - %P8)
Fiel
d Pe
rm.,
k x
10-5
cm
/sec
Gravel
Limestone
Figure 2.10 Field Permeability and Ratio of Coarse Sieve Difference (%P1/2” – %P3/8”) and Fine Sieve Difference (%P No.4 – %P No.8)
50
Fine Gradation, Coarse Sieve Delta
0.0
500.0
1000.0
1500.0
2000.0
2500.0
3000.0
0 5 10 15
Coarse Sieve Difference, %P1/2 - %P3/8
Fiel
d Pe
rm.,
k x
10-5
cm
/sec
Gravel
Limestone
Figure 2.11 Field Permeability and Coarse Sieve Difference (%P1/2” – %P3/8”)
Fine Gradation, Fine Sieve Delta
0.0
500.0
1000.0
1500.0
2000.0
2500.0
3000.0
0 5 10 15 20 25 30
Fine Sieve Difference, %P4 - %P8
Fiel
d Pe
rm.,
k x
10-5
cm
/sec
Gravel
Limestone
Figure 2.12 Field Permeability and Fine Sieve Difference (%P4 – %P8)
51
(4) Bailey Method. The Bailey Method provides a rational design method for
measuring and understanding the packing of aggregates (Vavrik et. al
2002). Three ratios were tested: (1) Coarse Aggregate Ratio, (2) Fine
Aggregate Coarse Portion Ratio, and (3) Fine Aggregate Fine Portion
Ratio (FAF). Only the latter ratio was found to be significant in explaining
changes in permeability. Computation for (FAF) are as follows:
19.0-mm NMAS mixes: FAF = %PNo.50 / %PNo.16.
12.5-mm NMAS mixes: FAF =.%PNo.100 / %PNo.30.
From a practical perspective, a larger relative percentage of material passing the
finer sieves (No.16 or No. 30) reduces mix permeability (see Figure 2.13).
However, there is a bell-shaped appearance in Figure 2.13 with all data and no
solid trend. When the gravel data were removed, there was a trend for limestone,
but there were low limestone ratios of 0.19 (USH-10 12.5-mm, Ndes=60 mix on
CABC) and 0.26 (Wisconsin Avenue 12.5-mm, Ndes=75 mix on PCC).
52
Fine Gradation, Bailey Method 3
0200400600800
100012001400
0.10 0.20 0.30 0.40 0.50
Bailey Method 3, Fine Aggregate Fine Portion Ratio
Fiel
d Pe
rm.,
k x
10-5
cm
/sec Gravel
Limestone
Figure 2.13 Field Permeability and Bailey Method 3 (Fine Mixes)
I. Passing No. 8 Sieve. Analysis of aggregate ratios found that coarse sieve and fine
sieve differences have an effect on permeability. Figure 2.14 illustrates the
relationship between three Passing No. 8 Sieve ranges (< 40%, 40-45%, and
>45%) against density and permeability. For similar density ranges, say 90% to
95%, the more coarse mixes (< 40% and 40-45%) were more permeable. This
finding is in agreement with the literature. Laboratory testing and analysis in the
later section of this report further addresses this issue.
53
Passing No. 8 Sieve and Permeability
0
1000
2000
3000
4000
5000
6000
7000
82 84 86 88 90 92 94 96 98
Core Density, %
Fiel
d Pe
rm.,
k x
10-5
cm
/sec
P8 < 40%
P8 = 40-45%
P8 > 45%
Figure 2.14 Field Permeability and Passing No. 8 Sieve
2.3.2 Coarse Mixes
The following are an interpretation of results for coarse mixes only. Due to a
limited data set and confounding of variables, several variables and their
interactions could not be tested for significance.
A. Base. Similar to fine mixes, two tests were conducted for base type: 2 levels
(rigid and flexible) and 3 levels (PCC, HMA, and CABC). In both tests, base had
a moderate impact on permeability. Figure 2.15 provides the relationship
between permeability and two base types, PCC and CABC. The STH-17 gravel-
source project constructed on CABC helped to produce a higher mean
permeability level.
54
Coarse Gradation, Base Type
0
1000
2000
3000
4000
5000
6000
7000
0 1 2 3 4
Base Type (1=PCC, 3=CABC)
Fiel
d Pe
rm.,
kx10
-5 c
m/s
ec
GravelLimestone
Figure 2.15 Field Permeability and Base Type (Coarse Mixes)
B. Density. For coarse mixes, there was moderate evidence that density affected
permeability, however, Figure 2.16 illustrates that this may be from higher
permeability on the more dense, gravel-source, STH-17 project. This figure also
shows no discernible trend between density and permeability for coarse mixes.
55
Coarse Gradation, Density
01000200030004000500060007000
80 85 90 95 100
Core Density, %
Fiel
d Pe
rm.,
kx10
-5 c
m/s
ec
GravelLimestone
Figure 2.16 Field Permeability and Density (Coarse Mixes)
C. Thickness. The plot in Figure 2.17 suggests that thicker layers are more
permeable, however, the moderate statistical significance of this variable was
attributed to the large scatter in the gravel-source data in the vicinity of 4 inches.
Figure 2.18 provides an important interactive relationship between density,
permeability, and thickness where density was more difficult to achieve on
thinner mats. Thus, thickness had an effect on achieving coarse-mix density, but
density had no effect on permeability.
56
Coarse Gradation, Thickness
01000200030004000500060007000
0 2 4 6 8
Core Thickness, inches
Fiel
d Pe
rm.,
kx10
-5 c
m/s
ec
GravelLimestone
Figure 2.17 Field Permeability Thickness (Coarse Mixes)
Coarse Gradation, Thickness and Density
0
1000
2000
3000
4000
5000
6000
7000
82 84 86 88 90 92 94 96 98
Core Density, %
Fiel
d Pe
rm.,
k x
10-5
cm
/sec
0-1 inch1-1.9 inch2-2.9 inch3-3.9 inch4-4.9 inch5-5.9 inch
Figure 2.18 Field Permeability, Thickness and Density (Coarse Mixes)
57
D. Thickness/NMAS Ratio. Smaller ratios for limestone-source mixes appeared
more permeable than higher ratios, as shown in Figure 2.19. No trend was
observed with the gravel-source mix. Figure 2.20 illustrates an interactive
relationship between t/NMAS ration, density, and permeability, where density
was more difficult to achieve with smaller t/NMAS ratios. Therefore, the
t/NMAS ratio had an effect on achieving coarse-mix density, however, density
had no effect on permeability. In general, t/NMAS ratios above 4 produced a
density above 92%.
Coarse Gradation, Thickness/NMAS Ratio
01000200030004000500060007000
0 2 4 6 8 10
Thickness/NMAS Ratio
Fiel
d Pe
rm.,
kx10
-5
cm/s
ec GravelLimestone
Figure 2.19 Field Permeability Thickness/NMAS Ratio (Coarse Mixes)
58
Coarse Gradation, t/NMAS and Density
0
1000
2000
3000
4000
5000
6000
7000
82 84 86 88 90 92 94 96 98
Core Density, %
Fiel
d Pe
rm.,
k x
10-5
cm
/sec
1-1.9 t/NMAS2-2.9 t/NMAS3-3.9 t/NMAS4-4.9 t/NMAS5-7.9 t/NMAS
Figure 2.20 Field Permeability, Thickness/NMAS Ratio and Density (Coarse Mixes)
2.3.3 Density Growth
Analysis of variance was conducted for density growth on individual projects
with results shown in Tables 2.4 and 2.5. Number of passes were adjusted in the analysis
to compare results of initial density gain with the breakdown roller with density growth
across all rollers. First, all passes were used (Table 2.4); then only the first 4 passes were
used to assess initial gain in the density (Table 2.5).
Project data were not pooled since different equipment and materials were used
among projects, and pooled results would have been difficult to interpret and generalize
across similar variables. For example, a thin pavement compacted with low lab air voids
may be easy to compact primarily because of the low lab air voids, however, a statistical
significance test may possibly yield the thin layer as the significant variable. Thus, the
analysis “blocked” and removed project variables such as mix type, %AC, % dust, lab air
59
voids, VMA, roller weight, roller width, and numerous other project factors. By
removing those variables, the analysis was then able to focus on key independent
variables of interest though to affect density growth, namely, layer thickness, number of
passes, mat temperature, and their interactions.
60
Table 2.4 Statistical Significance Results for Density Growth (All Passes)
I-43 STH-23 STH-23 Wis Ave Wis Ave USH 10 I-894 I-894 STH-21 STH-21 USH-8 USH-8 USH-8 I-94 STH-17 19-mm 19-mm 12.5-mm 19-mm 12.5-mm 19-mm 19-mm 12.5-mm 19-mm, 12.5-mm, 19mm(1) 19mm(2) 12.5-mm 25-mm 25-mm
Variable Coarse Fine Fine Fine Fine Fine Fine Fine Fine Fine Fine Fine Fine Coarse Coarse (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)
Degrees of Freedom 127 100 87 113 59 58 108 81 96 233 132 157 71 106 92 Thick N/S ** *** N/S N/S *** N/S N/S *** *** *** N/S N/S *** *** Temp --- *** *** *** *** *** *** *** *** *** *** *** *** *** *** Thick*Temp --- * N/S ** N/S N/S N/S *** N/S *** N/S ** ** N/S *** Passes *** N/S N/S *** *** *** *** *** *** *** *** ** *** *** N/S Thick*Pass N/S N/S N/S ** N/S N/S N/S N/S * N/S N/S N/S *** N/S N/S
Temp*Pass --- *** *** *** *** *** *** *** N/S *** *** *** *** *** *** Thick*Temp*Pass --- N/S *** ** N/S N/S N/S ** N/S *** ** N/S N/S N/S N/S Density Testing,% 48 22 15 32 22 10 9 13 32 18 16 29 10 16 21
Significance Levels: N/S = Not Significant; * = 0.05 < p-value < 0.10;** = 0.01 < p-value< 0.05; *** = p-value < 0.01
61
Table 2.5 Statistical Significance Results for Density Growth (First 4 Passes)
I-43 STH-23 STH-23 Wis Ave Wis Ave USH 10 I-894 I-894 STH-21 STH-21 USH-8 USH-8 USH-8 I-94 STH-17 19-mm 19-mm 12.5-mm 19-mm 12.5-mm 19-mm 19-mm 12.5-mm 19-mm, 12.5-mm, 19mm(1) 19mm(2) 12.5-mm 25-mm 25-mm
Variable Coarse Fine Fine Fine Fine Fine Fine Fine Fine Fine Fine Fine Fine Coarse Coarse (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)
Degrees of Freedom 38 45 43 72 36 49 55 31 35 100 89 53 17 26 22 Thick N/S ** *** N/S N/S *** *** *** *** N/S *** N/S N/S N/S *** Temp --- *** *** *** *** *** *** *** *** *** *** *** *** *** *** Thick*Temp --- *** *** *** N/S * N/S N/S N/S *** N/S ** * ** *** Passes *** *** *** *** *** N/S *** *** *** *** *** *** *** *** *** Thick*Pass ** ** *** ** *** * ** N/S N/S *** N/S * N/S * * Temp*Pass --- N/S *** N/S *** *** *** *** *** *** *** *** N/S *** N/S Thick*Temp*Pass --- N/S * *** * N/S N/S N/S ** *** N/S N/S N/S N/S *** Density Testing, % 9 8 6 20 16 10 11 7 8 19 8 22 4 17 4
Significance Levels: N/S = Not Significant; * = 0.05 < p-value < 0.10;** = 0.01 < p-value< 0.05; *** = p-value < 0.01
62
In Table 2.4 (all passes), it was determined that mat temperature was found to be
significant across projects and different NMAS layers. Number of passes was significant
on all limestone-source projects, while 2 of 3 gravel-source projects did not find passes to
be significant. This is supported by Figure 2.5 where fine gravel-source projects had a
slightly higher final density than fine limestone-source projects. The interaction of
passes with the declining mat temperature was significant on all projects, except the
STH-21 19-mm mix.
During field data collection, it was observed that a greater density gain was
achieved during the initial passes with the breakdown roller, then decreasing mix
temperature and void spaces increased the resistance of the mat to densification. In Table
2.5, the data set was reduced to the first 4 passes to analyze densification with the
breakdown roller. Mat temperature was again found to be a significant factor in density
growth. The interactions of thickness and passes, and thickness and temperature were
also influential on a majority of project layers. Similar to the full-pass analysis, thickness
was not a significant factor in density growth across all projects during initial breakdown
compaction.
Plots of the changes in density mean levels with varying thickness for I-894 are
provided in Figures 2.21 and 2.22. These growth trends are typical of all projects where
there was a relatively large density increase from initial passes, then a tapering effect
with remaining passes. Final growth was as a function of passes, and the effect of
thickness was random. Additional project density-growth plots are provided in Appendix
B. Due to various field constraints, it was not possible to collect data on each project
layer in the study.
63
I-894 Lower Layer Density Growth
70
75
80
85
90
95
0 5 10 15 20
Roller Passes, n
Den
sity
, %
1.875in2.75in (2)2.75in (3)2.875in3.25in (6)3.75in
Figure 2.21 Density Growth on I-894 19-mm Lower Layer Mix
64
I-894 Surface Layer Density Growth
70
75
80
85
90
95
100
0 5 10 15
Roller Passes, n
Den
sity
, %
1.625in1.75in1.875in2in2.5in (4)2.5in (6)
Figure 2.22 Density Growth on I-894 12.5-mm Surface Layer Mix
Based on the statistical analysis, and representative plots from the I-894 project,
density growth was primarily influenced by mat temperature and number of passes.
Layer thickness was a factor on a project-specific basis.
2.4 Investigation of Specification Criteria
Density is one of the most important factors of a durable, long-lasting pavement.
This study found that lower density, limestone source, fine-graded Ndes=100 mixes were
more permeable, while no trend was observed for gravel-source mixes. NCAT has
recommended the following permeability criteria for coarse-graded mixes only, as shown
in Table 2.6. To date, no other coarse-graded recommended criteria have been published,
nor have any criteria been published for fine-graded mixes. These criteria are based upon
65
the NMAS of the mix, and critical values should likely be different for fine- and coarse-
graded mixes even though the NCAT study involved only coarse-grades mixes (Cooley
et. al 2001).
Table 2.6 NCAT Recommended Permeability Criteria for Coarse-Graded Mixes (Adopted from [1])
NMAS, mm
(1)
K x 10-5 cm/sec, maximum
(2)
%Gmm, minimum
(3) 9.5 100 92.3 12.5 100 92.3 19 120 94.5 25 150 95.6
As stated earlier, NCAT and other studies have determined NMAS has a
significant effect on permeability, while this study has not found NMAS to be a
significant factor. Thus, applying NMAS-specific criteria would not be justified.
However, for sake of investigation, an analysis was conducted to determine if NCAT
criteria are achievable. Table 2.7 provides both a frequency and percentage of all fine-
graded test sites that met, or failed to meet, these permeability and density thresholds.
Field permeability was measured using the NCAT permeameter, and density was
measured using field cores. For nearly all criteria combinations, a greater percentage of
failure occurred.
Table 2.7 Pass/Fail Results for Fine-Mix WisDOT Projects applied to NCAT Recommended Permeability Criteria
Criteria (1)
K x 10-5 cm/sec (2)
%Gmm (3)
(a) 12.5-mm NMAS
66
Pass 18 40% 23 51% Fail 27 60% 22 49% Total 45 100% 45 100%
(b) 19-mm NMAS Pass 23 41% 5 9% Fail 33 59% 51 91% Total 56 100% 56 100%
The preferred method to establish a minimum density and maximum permeability
value for various mix classifications is with actual performance data. This study can
provide recommendations for the level of density needed to control permeability,
however, the level of permeability to achieve a durable, long-lasting pavement is not
known (for example, should the minimum level be k = 100x10-5 cm/sec, k = 300x10-5
cm/sec, or some other value). Both short-term and long-term monitoring of pavement
performance is necessary to establish true levels.
Until a performance-based approach is adopted, a beginning approach would be to
find the median value that produces a 50-50 pass-fail percentage. Table 2.8 provides
cross-classified permeability values for all data, by source, and by Ndes. Best-fit
regression equations to determine required density are also provided.
Table 2.8 Median “k” Values and Density for Fine-Mix Projects
Level (1)
Number of Samples
(2)
k x 10-5 cm/sec
(3)
Regression Equation
(4)
R2 (5)
Density Required,
% (6)
(a) All Data - - - 101 152 Perm. = 10185 -
106.9xDensity 17.1 93.8
(b) Source Limestone 62 220 Perm. = 17525 -
185.5xDensity 31.1 93.3
Gravel 39 114 Regression Equation Not Significant
--- ---
67
(c) Ndes 60 12 28 Regression Equation
Not Significant --- ---
75 53 97 Regression Equation Not Significant
--- ---
100 36 627 Perm. = 18167 -190.0xDensity
32.6 92.3
One aggregate ratio has the potential to control permeability of fine-grades mixes:
(%P1/2 - %P3/8) / (%PNo.4-%PNo.8). This ratio will be referred to as the “Coarse Sieve
Delta / Fine Sieve Delta Ratio”. Higher ratios have the potential to reduce permeability.
Higher permeability results as the gaps increase between coarse aggregates (%P1/2” and
%P3/8”) and fine aggregates (%P4 and %P8), respectively. This suggests relative
differences in these sieves may have an effect on internal void structure, and measured
permeability, of the compacted material. The mix design would be adjusted to a
sufficient ratio to withstand permeability by either increasing the difference between the
coarse sieves, reducing the difference between the fine sieves, or both.
Table 2.9 provides sample data and calculations for this ratio. In the first example
using data from I-894 (12.5-mm, E-30, limestone-source mix) specifies k=152 x10-5
cm/sec for all fine mixes, k=220 x10-5 cm/sec for limestone source, and k=627 x10-5
cm/sec for E-30, Ndes=100 mixes. Based on the fine mix permeability threshold (k=152
x10-5 cm/sec), a ratio exceeding 1.00 is desired. For the source permeability threshold
(k=220 x10-5 cm/sec), a ratio of 0.8 is needed. Finally, for the Ndes classification (k=627
x10-5 cm/sec), a ratio of 0.8 should ensure this criteria is met.
In the second example using data from STH-23 (19-mm, E-3, gravel-source mix),
gravel sources are more robust to the ratio. This mix would specify k=152 x10-5 cm/sec
68
for all fine mixes, k=114 x10-5 cm/sec for gravel source, and k=97 x10-5 cm/sec for E-3,
Ndes=75 mixes.
Table 2.9 Mix Design Calculations for Coarse Sieve Delta / Fine Sieve Delta Ratio
Mix Design
(1) % Passing ½”
(2) % Passing 3/8”
(3) % Passing No.4
(4) % Passing No.8
(5) I-894, 12.5-mm 98.4 84.4 62 41.3 Coarse Delta = 14.0 Fine Delta = 20.7 Coarse Delta / Fine Delta = 14.0/20.7 = 0.68 STH-23, 19-mm 90.3 84 68.9 51.2 Coarse Delta = 6.3 Fine Delta = 17.7 Coarse Delta / Fine Delta = 6.3/17.7 = 0.36
The data analysis found no discernible trend between density and permeability for
coarse mixes. As described earlier, NCAT has published recommended criteria for
coarse-graded Superpave mixes. An analysis was conducted to determine if these criteria
are achievable. Table 2.10 provides a frequency and percentage of all coarse-graded test
sites that met, or failed to meet, recommended NCAT permeability and density
thresholds. For all criteria combinations, a large percentage of failure occurred. Less
than 20% of the criteria were met.
Table 2.10 Pass/Fail Results for Coarse-Mix WisDOT Projects applied to NCAT Recommended Permeability Criteria
Criteria (1)
K x 10-5 cm/sec (2)
%Gmm (3)
(a) 9.5-mm NMAS Pass 5 83% 1 17% Fail 1 17% 5 83% Total 6 100% 6 100%
(b) 19-mm NMAS Pass 1 6% 1 6%
69
Fail 16 94% 15 94% Total 17 100% 16 100%
(c) 25-mm NMAS Pass 0 0% 2 13% Fail 15 100% 14 87% Total 15 100% 16 100%
Similar to fine mixes, a beginning approach to establish a maximum permeability
level would be to calculate the median value that produces a 50%/50% pass/fail
percentage. Table 2.11 provides cross-classification for all pooled data, source, and
Ndes. Data from the USH-20 Rockford, Illinois Ndes=70 mix was combined with the
Ndes=75 data from STH-17. Best-fit regression equations are provided to determine
required density.
Table 2.11 Median “k” Values for Coarse-Mix Projects
Level (1)
Number of Samples
(2)
k x 10-5 cm/sec
(3)
Regression Equation
(4)
R2 (5)
Density Required,
% (4)
(a) All Data - - - 38 913 Regression Equation
Not Significant --- ---
(b) Source Limestone 31 718 Perm. = 11950 -
120.2xDensity 17.8 93.5
Gravel 7 2,790 Perm. = 164029 -1694.4xDensity
66.2 95.2
(c) Ndes 75 24 1,100 Regression Equation
Not Significant --- ---
100 6 199 Perm. = 7718 -80.4xDensity
23.6 93.5
125 8 1,560 Regression Equation Not Significant
--- ---
70
2.5 Summary of Findings from Field Study
Table 2.12 summarizes permeability and density results from the field study.
Based on the fine-graded mix data, higher Ndes limestone-source mixes were more
permeable. It was also determined that t/NMAS ratio was influential in achieving density
below a ratio of 2 for gravel sources and ratio of 3 for limestone sources. A “tapering
effect” was observed for limestone-source mixes outside the current WisDOT thickness-
to-NMAS range of 3 to 5, where it was more difficult to achieve density below a ratio of
3, and possible to achieve a 92% density above a ratio of 5. Limestone-source fine mixes
were more difficult to compact on a rigid PCC base. There were several factors that
affected density growth, including mat temperature, number of passes, and their
interaction. Layer thickness was a factor on a project-specific basis, with some projects
indicating it was significant and others not significant.
Table 2.12 Summary of Field Study
Variable (1)
Fine-Graded Mixes (2)
Coarse-Graded Mixes (3)
Base More difficult to compact on PCC base and higher Ndes-level mixes, yielding a more permeable pavement.
CABC was more permeable; only one project constructed on CABC.
Source Limestone source was more permeable for high Ndes=100 mixes.
Confounded data; no statistical determination possible.
Density Limestone source affects density, but no trend was observed for gravel.
No effect on permeability.
Ndes Higher Ndes levels were more permeable.
Confounded data with no determination possible.
Thickness Limestone sources had lower density than gravel for similar thickness. Thickness was an inconsistent factor affecting density growth.
Inconsistent factor for density growth.
NMAS No effect on density and permeability.
No effect on density and permeability.
71
Passing No. 8 Sieve Fine mixes with 40-45% Passing No. 8 sieve were more permeable than >45% Passing No. 8 sieve.
Coarse mixes with <40% Passing No. 8 sieve were more permeable than fine mixes with >40% Passing No. 8 sieve.
Thickness/NMAS Limestone with lower ratios was less permeable. Gravel had no trend.
No effect on density and permeability.
Aggregate Ratios A small Coarse Delta / Fine Delta Ratio was more permeable.
No determination made.
Mat Temperature Higher temperatures help achieve greater density gain.
Higher temperatures help achieve greater density gain.
Number of Passes More passes help achieve greater density gain.
More passes help achieve greater density gain.
One aggregate ratio has the potential to control permeability of fine-grades mixes:
Ratio of (%P1/2 - %P3/8) / (%PNo.4-%PNo.8), referred to as the “Coarse Sieve Delta /
Fine Sieve Delta Ratio”. High ratios produced lower permeability. In addition, higher
permeability results as the gaps increase between coarse aggregates (%P1/2” and
%P3/8”) and fine aggregates (%P4 and %P8), respectively. This suggests relative
differences in these sieves may have an effect on internal void structure, and measured
permeability, of the compacted material. The mix design could be controlled to a
sufficient ratio to withstand permeability by either reducing the difference between the
coarse sieves, fine sieves, or both.
For both fine and coarse mixes, an interactive relationship was found between
density, permeability, and thickness where density was more difficult to achieve on
thinner mats, however, density had no little effect on permeability for fine mixes and no
effect on coarse mixes.
Based on limited coarse-graded data collected in this study, a clear relationship
between layer thickness and permeability was not established. For the thickness/NMAS
72
ratio, smaller ratios for limestone-source mixes appear more permeable than higher
ratios, and no trend was observed with the gravel-source mix.
73
CHAPTER THREE
LABORATORY DATA ANALYSIS AND DISCUSSIONS
3.1 Introduction
This chapter provides details of the laboratory testing procedures used in this
project, the results collected, and the data analysis followed. The lab testing procedures
include the measurement of density and permeability of field cores, the alternative for the
SGC specimen compaction methods, and the proposed procedure for testing the
permeability of the lab compacted specimen and estimating the permeability of the field
specimen. The results of the study include the lab density and permeability of field cores,
and the lab density and permeability of SGC specimens. Based on the testing results, the
analysis such as, correlation between field density and lab density, correlation between
field permeability and lab permeability, and correlation between lab permeability of field
cores and SGC specimens, are covered in this chapter.
3.2 Field Cores Permeability Testing
3.2.1 Equipment and Methods
In the laboratory, field cores and loose mix samples were taken from field
projects. The cores were used to measure the density using the vacuum sealing method by
the Corelok Device. All field cores were completely dried at room temperature before
measuring the density with the Corelok. The loose mix samples were used to measure the
maximum specific gravity (Gmm). It was determined that the Gmm measured in the lab
74
using the Corelok is essentially similar to the Gmm measured in the plant using the
ASTM D 2041.
To measure laboratory permeability of field cores, a flexible-wall permeameter
and a pressure panel board following the methods described in ASTM D 5084-01
(Standard Test Method for Measurement of Hydraulic Conductivity of Saturated Porous
Materials Using a Flexible-Wall Permeameter), was used. A flexible-wall permeameter
used in ASTM D 5084 is shown in Figure 3.1. The specimen is placed between two caps
(upper and lower caps) in a cell filled with water. A latex membrane is used to seal the
specimen to the caps and to isolate the specimen from the water in the cell. Tubing is
routed to the upper and lower caps for flowing water through the specimen. Water in the
cell is used to apply stress to the specimen and to ensure that the membrane remains in
tight contact with the specimen. Tight contact is critical because it prevents flow along
the interface between the specimen and the membrane.
A pressure panel board is attached to the permeameter for delivery of water at
specified pressures and for measuring the rate at which the water flows. A pressure panel
typically consists of at least three burettes and a variety of regulators and valves for
distributing the water and controlling the applied pressures. A schematic of a pressure
panel is shown in Figure 3.2. One burette is to measure volume changes that occur
within the cell. The other two burettes are used to measure the rate of flow into and out
of the specimen. The regulator associated with each burette is used to control air pressure
applied on top of the water in the burettes, and thus controls the water pressure in the cell
or the inflow and outflow lines.
75
Before the specimen is permeated, it is saturated using a technique called “back-
pressuring”, which consists of incrementally increasing the cell pressure and the influent
and effluent pressure in equal amounts until the specimen becomes saturated. The
elevated pore water pressure in the specimen (caused by the applied “backpressure” at the
influent and effluent ends) forces water into small air-filled pores, collapses air bubbles,
and enhances the rate at which air bubbles diffuse into the water. By incrementing the
cell pressure and pore water pressures in equal amounts, the net stress (cell pressure-pore
water pressure) acting on the specimen is unchanged. This stress is referred to as the
“effective stress” in geotechnics.
The check for saturation during back-pressuring uses Skempton’s B coefficient
for particulate materials (Skempton 1954). Skempton’s B coefficient is the ratio of the
rise in pore water pressure (∆u) relative to an incremental change in the cell pressure
(∆pc) when the valves on the inflow and outflow lines (Figure 3.2) are closed:
cp
uB∆∆
= (3.1)
In theory (Skempton 1954), an increase in the cell pressure will result in equal change in
pore water pressure in a particulate specimen that is saturated and from which drainage is
prevented; i.e., the B coefficient will be 1. This condition prevails since the solids and
water are essentially incompressible, the water is a continuum, and the solids are
individual particles. In reality, particle-to-particle contacts preclude the B-coefficient
from ever reaching 1.0 (Bishop and Eldin 1950) and specimens with B ≥ 0.95 are
generally accepted as saturated (Daniel 1994).
Cementation, as in HMA, can further reduce the B-coefficient at saturation. For
such materials, a reasonable threshold for B at saturation can be assessed by measuring
76
the B-coefficient daily, followed by a concurrent and equal increment in the backpressure
and cell pressure (typically the increment is 35 kPa). At some backpressure, the B-
coefficient will cease increasing, and this threshold value corresponds to saturation. No
general threshold value for B has been established for HMA. Thus, based on the initial
testing results, the common criterion of B ≥ 0.95 appears reasonable for testing HMA.
The hydraulic gradient (i) is applied after the specimen has been saturated through
back-pressuring. The hydraulic gradient is the ratio of the drop in total head across the
specimen (∆H) to the length of the specimen (L); i.e., i = ∆H/L. Several methods can be
used to apply the hydraulic gradient; they differ in how the difference in total head across
the specimen is controlled. The methods in D 5084 are (i) the constant head method
(Method A), (ii) the falling-head method (Method B), and (iii) the falling-head rising-tail
method (Method C). In this study, the falling head-rising head method (D 5084-Method
C) was used since it is easily implemented while applying backpressure to the specimen.
In this method, the total head on the influent end of the specimen decreases while the
total head on the effluent end increases. Thus, the difference in total head across the
specimen decreases during the test. The changes in the drop in total head are read
directly off the burettes as changes in the water levels. The permeability (K) is computed
using the following (Daniel 1989):
)HH
ln())tt(A
L)(aa
aa(K
2
1
1221
21
∆∆
−+= (3.2)
where a1 and a2 are the cross-sectional area of inflow and outflow burettes, respectively,
A is the cross-sectional area of the specimen, and ∆H1 and ∆H2 are the differences in total
head across the specimen at times t1 and t2 respectively.
77
Ideally the hydraulic gradient should be selected as close as possible to the value
expected in the field. For pavements, the gradient is likely to be close to one since
appreciable ponding on the surface does not occur and gravity is the primary mechanism
driving flow. However, testing times can be long when a hydraulic gradient near one is
used, especially for less permeable materials. ASTM D 5084 provides guidelines on
maximum values for the hydraulic gradient that depend on the anticipated permeability.
However, these guidelines were developed for soils, which are softer than HMA and thus
are more susceptible to compression caused by seepage pressures. Based on the initial
results of the permeability testing of HMA, an intermediate hydraulic gradient of 18 was
used for this study.
During the test, water is allowed to flow through the specimens for three times,
and hence, three permeability readings are measured at the steady state. The average
value of the three permeability readings is reported. However, it should be noted that
duplicate testing, using two different specimens of the same mix, is not conducted
because of the difficulty in obtaining duplicate specimens that has the exact same density
and distribution of the voids.
78
Vent Port
Top Plate
"O" Ring Seal
Top CapAcrylic Cylinder
Specimen
"O" Ring Seal
Latex Membrane
"O" Ring Seal
Bottom Plate
EffluentInfluent
Bottom Flush
Top Flush
Cell PressureLine
Base Pedestal
Porous Disks
Flexible Tube
Figure 3.1 Flexible-Wall Permeameter Used in ASTM D 5084
79
Cell EffluentInfluent
AirPressureRegulator
Air-Water
Interface77.32
PressureReadout
WaterValves
TubingConnectors
Air
Tube toCell
Tube toInfluent
Tube toEffluent
Water
Burette
Figure 3.2 Pressure Panel Used in ASTM D 5084
80
3.2.2 Density and Permeability Results
Table 3.1 includes a summary of the field and laboratory results for density and
permeability. The laboratory results are for the field cores specimens, not the specimens
compacted in the laboratory using the SGC, which will be covered in a later section.
Table 3.1 Summary of All Test Results
Field Data Field Cores Data Project
Site
Layer
Thickness (cm)
Field Density (Nuclear Gauge)
(%)
Field Permeability
(NCAT Device) (x 10–5 cm/s)
Lab Density (CoreLok Device)
(%)
Lab Permeability (ASTM D5084)
(x 10–5 cm/s)
STH23U
STH23L
WiscU
WiscL
USH110U
USH110L
USH8U
1 2 3 4 5 6
1 2 3 4 5 6
1 2 3 4 5 6
1 2 3 4 5 6 7 8 9 10 11
1 2 3 4 5 6
1 2 3 4 5 6
1 2 3 4 5
3.5 4.3 5.5 5.9 6.1 6.5
7.9 8.1 4.2 5.2 3.4 3.1
3.8 4.1 6.0 3.2 6.4 5.1
7.9 7.0 6.4 5.1 4.5 5.7 5.1 3.8 3.2 3.5 3.2
4.8 5.1 3.5 3.8 3.5 3.2
2.9 3.5 4.5 4.8 7.3 7.0
6.0 5.4 6.0 5.7 3.8
93.77 93.42 93.20 93.24 92.26 91.54
91.10 91.38 90.63 92.00 90.22 90.20
92.25 94.38 90.63 90.43 92.24 94.12
94.62 90.68 93.31 90.77 92.40 90.87 83.22 86.27 85.36 89.94 90.88
92.83 91.60 92.07 93.51 92.90 91.83
93.03 95.20 91.77 92.37 92.10 91.33
91.44 92.16 92.73 92.45 91.48
8.23 18.81 11.33 46.07 46.45
115.91
21.02 22.49 64.81 48.73 77.97
128.65
34.88 26.34
220.14 171.06 39.10 22.46
58.05 38.64
186.06 399.32 21.83
140.10 2664.92 1547.70 2491.49
62.57 82.02
50.11 28.08 27.39 55.19 15.86 31.21
5.32 6.36 14.94 36.50 53.10
104.73
318.46 243.52 153.89 172.43 129.15
92.71 91.76 93.66 93.54 92.51 92.24
91.84 92.47 90.58 92.00 90.35 87.94
91.47 93.88 91.82 90.87 92.57 94.15
95.69 92.37 92.99 91.43 93.58 92.41 85.64 86.42 84.50 91.70 92.29
92.23 92.35 92.03 92.35 93.90 92.83
92.67 93.02 92.76 92.70 93.38 92.79
91.56 91.81 92.94 92.31 91.72
25.40 36.00 21.70 34.20 47.90 69.13
35.27 31.35 41.67 32.97 59.97 62.20
38.45 16.55 52.73 50.35 19.50 9.91
0.09 34.30 45.00 46.85 9.75 40.70 93.85 70.85 79.10 46.03 26.35
30.75 24.4 27.95 27.75 5.53 22.65
5.91 7.27 20.65 24.40 23.95 20.70
71.9 71.7 40.3 54.3 50.9
81
USH8L
I894U
I894L
STH21U
STH21L
I94
USH20U
USH20L
STH17
6
1 2 3 4 5 6
1 2 3 4 5 6
1 2 3 4 5 6
1 2 3 4 5 6
1 2 3 4 5 6
1 2 3 4 5 6
1 2 3 4 5 6
1 2 3 4 5 6
1 2 3 4 5 6
4.1
3.8 4.1 4.8 5.4 6.4 5.7
4.8 4.1 6.4 6.4 4.4 5.1
4.8 7.0 7.0 7.3 9.5 8.3
5.7 5.4 6.4 5.7 4.1 4.4
7.0 7.6 7.6 7.0 7.0 5.1
10.2 10.2 11.8 12.1 7.0 7.0
3.5 4.7 4.6 4.1 3.5 3.4
2.2 2.2 4.4 4.4 6.4 7.0
9.5 9.5
11.8 11.1 11.1 11.8
92.36
93.04 93.25 93.86 94.28 93.71 93.43
94.51 94.82 92.62 91.83 89.33 92.63
90.91 91.75 93.18 93.50 93.45 89.50
92.40 91.94 92.06 93.29 91.86 91.30
91.74 90.94 91.47 92.73 91.74 91.72
93.86 91.60 93.24 92.27 91.60 93.43
88.95 93.48 92.59 91.95 93.10 92.91
91.50 91.97 92.21 90.53 92.12 90.90
90.07 91.28 92.03 91.70 92.81 90.52
99.16
116.19 151.06 145.25 151.95 315.97 220.13
208.3 98.9 571.5 725.4
1880.5 902.6
2006.65 483.95 876.03 235.10 627.15 2586.68
520.0 778.2 693.3 564.9 172.6 273.8
1014.29 975.28 954.99 369.59 333.25 155.48
1560.0 2860.1 793.8
1567.5 694.0 167.1
1465.9 557.0 208.5 136.2 75.2 103.1
976.4
2114.1 913.0
1337.1 1816.1 3325.9
6023.52 1099.18 597.50 3322.09 2789.17 11187.43
93.18
92.85 93.57 93.08 94.02 94.10 93.88
94.15 93.61 92.35 92.94 87.60 91.37
90.33 92.66 93.20 93.43 94.33 91.14
92.52 91.71 92.43 92.57 92.57 90.65
89.86 90.58 90.60 91.86 91.38 91.29
93.18 92.04 93.65 91.76 91.17 92.12
87.54 93.73 92.97 93.73 93.37 93.61
83.53 82.46 90.01 89.10 90.24 89.18
94.11 96.23 96.70 96.86 96.54 94.62
37.1
100.3 49.7 51.6 24.3 35.4 40.8
25.33 56.60 55.70 49.30 97.97 62.83
91.00 41.73 72.27 12.00 0.41 72.13
65.1 96.5 81.3 84.7 37.9 73.9
70.1 39.4 90.8 108.0 105.0 90.3
33.30
103.10 27.23 31.47 82.97 9.0
83.10 8.75 53.0 40.9 25.9 26.8
48.8 29.6 27.3 93.2 56.6 88.8
35.2 3.83 1.27 0.76 2.52 57.0
82
3.3 Laboratory Compacted Specimen Testing
The laboratory work also included production of compacted mixture samples
from loose mixtures using the Superpave Gyratory Compactor (SGC). The objectives
were to better mimic aggregate orientation in the field, and to develop a tentative
procedure to estimate permeability of field pavements. To achieve both of these
objectives, two compaction procedures were evaluated. The following sections
summarize the compaction procedure, the compaction results, and the permeability
results of the SGC specimens produced by these compaction procedures. This section
also includes recommendations to validate a protocol to estimate permeability in the lab
of mixtures intended for construction in the field after completing the mix design.
3.3.1 Alternatives for Laboratory Compaction Method
Initially, the compaction procedure, called here “Method A”, was selected to
produce the SGC specimens that have the same air voids content as the field cores. It
was assumed that because voids content is a major factor controlling permeability, the lab
permeability of these specimens would be very similar to the field permeability. Method
A is a trial and error procedure that can be summarized as follows:
Method A
After the field cores were taken to the laboratory, height (thickness) and Gmb of
field cores were measured. The amount of loose asphalt mixtures was then calculated to
produce SGC specimens that will have the same height and density at the specified
number of gyrations. The equation to calculate the amount of material for the
compaction is as follows:
83
Wt. = Gmb x t x A / 1000 (3.3)
where Wt. is the amount of material (g), t is the height or thickness of field cores (mm),
and A is the cross sectional area of the specimen (mm2). The SGC specimens were
compacted by controlling specimen height in the Superpave gyratory compactor. The
Gmb was measured after the curing of compacted specimen for 24 hours at room
temperature. If the density varies from the targeted density by ± 0.5%, another SGC
specimen will be compacted by adjusting the specimen height until the targeted density is
achieved at the field core thickness (height). The compaction by Method A resulted in
the SGC specimens with approximately the same density and thickness as the field cores.
However, the number of gyrations had to be varied depending on the mix type or source,
the targeted density and thickness. The ASTM D5084 was used to measure the lab
permeability of these SGC specimens.
Figure 3.3, shows the relationship between the lab permeability of the specimens
produced by Method A using the SGC compactor and the permeability of field cores
measured by the same method in the lab. As shown in Figure 3.3, the permeability of the
SGC specimens are generally lower than the field cores and there is very high scatter in
the results indicating the lack of good correlation between the two measures. To
determine the cause of difference between lab permeability of field cores and lab
permeability of SGC specimens, Method B was proposed to evaluate hypothesis that the
difference in the lab permeability of field cores and the permeability of SGC samples are
due to different aggregate orientation which could be resolved by changing amount of
material placed in mold as explained by Dr. Erv Dukatz of Mathy Construction.
84
0
20
40
60
80
100
0 20 40 60 80 100
Lab
Perm
eabi
lity
of F
ield
Cor
es (x
10-5
cm
/s)
Lab Permeability of SGC Specimen (x 10-5 cm/s)
y = 22.24 + 0.734x
R2 = 0.60
Figure 3.3 Relationship between lab permeability of field cores and SGC specimens.
One project (the STH23-lower mix) was selected as a trial mix for the compaction using
Method B. The compaction using Method B can be summarizes as follows:
Method B
In this method, the loose asphalt mixture was used to compact different sample
sizes but using the same number of gyrations. At least three specimen sizes were selected
based on the range of field cores density and thickness. The number of gyrations was
fixed at Ndes = 75 gyrations. The compaction by Method B results in the SGC
specimens within the range of same density as in the field cores but varied in the
specimen thickness. Similar to specimens produced by Method A, the ASTM D5084 was
used to measure the permeability for SGC specimens. The lab permeability of these SGC
specimens was found to be close to the lab permeability of field cores and to be within
85
the same range of density. Figure 3.4 shows the comparison of the lab permeability
between the field cores and the SGC specimens compacted by using Method A and
Method B for the selected project. The results show that Method B, which is a
compaction at fixed number of gyrations at Ndes, is giving the best alternative for the
compaction to simulate the field specimens.
0
10
20
30
40
50
60
70
86 88 90 92 94 96
Lab
Perm
eabi
lity
(x 1
0-5 m
/s)
Density (%)
Field Cores
Method B
Method A
Figure 3.4 Lab permeability of field cores and SGC specimens compacted by Method A and Method B (STH23-Lower Mix)
3.3.2 Proposed Compaction Procedure
Based on the comparison of the results for Methods A and B, method B was used
for the laboratory study in the remaining of this project. The detailed steps of the
procedure includes 4 steps, as follows:
86
Step 1: Based on the mix design process, the loose asphalt mixture (field-mixed) was
obtained from particular field project. In this study, the loose mix collected from field
study as indicated in Chapter 2 was used.
Step 2: At least three SGC specimens were compacted to Ndes, which is depending on
the mixture type. The amount of material used to produce different sizes of specimens
was in the range of 1000 to 4000 g.
Step 3: Density (%Gmm) and lab permeability of these specimens were measured after
the compaction was completed. Figure 3.5 shows an example for the plot of density
versus lab permeability of SGC specimens.
0
10
20
30
40
50
60
89 90 91 92 93 94 95
Perm
eabi
lity
(x 1
0-5 c
m/s
)
Density (%)
Compact at Ndes
Figure 3.5 Density vs. Lab Permeability
87
Step 4: Since the targeted field density is known, the permeability can be predicted
based on Figure 3.5. For example, if the targeted field density is 92% Gmm, the
predicted permeability in the field is approximately 32 x 10-5 cm/s.
The procedure was used for several other projects included in this study. The
predicted permeability values are plotted versus the measured permeability of field cores
as shown in Figure 3.6. The relationship between predicted and measured permeability
confirm that the procedure show a very good potential.
To further validate the proposed procedure, a total of 16 mixes as listed in
Chapter 2 were compacted and the permeability was measured as described in Step 1 to
3. It should be noted that the specimen thickness is considered as a secondary factor,
because it is normalized in the analysis of the permeability test results. Therefore, the
density, which is a main factor affecting permeability, was targeted to estimate the
permeability.
88
20
30
40
50
60
70
20 30 40 50 60 70
Pred
icte
d Fi
eld
Perm
eabi
lity
(x10
-5 c
m/s
)
Measured Field Permeability (x10-5 cm/s)
Figure 3.6 Predicted vs. Measured Field Permeability
3.3.3 Density and Permeability Results
The density and the lab permeability results of all SGC specimens that were produced
and tested according to Method B are summarized in Table 3.2. The table is organized
according to project. As can be seen in the table, the projects included various Ndes
values, multiple thickness and density values. For each project a minimum of 2 SGC
samples of different thickness was produced and used to predict permeability. The
relationship between the field and laboratory density and field and laboratory
permeability are discussed in section 3.4 as part of the analysis.
89
Table 3.2 Summary of density and lab permeability results of all SGC specimens
Field Cores SGC Specimens
Project Ndes Thickness
(cm)
Density (%) Permeability
(x 10–5 cm/s)
Thickness
(cm)
Density (%) Permeability
(x 10–5 cm/s)
STH23U 75 3.5 4.3 5.5 5.9 6.1 6.5
92.71 91.76 93.66 93.54 92.51 92.24
25.40 36.00 21.70 34.20 47.90 69.13
3.7 4.2
92.38 93.77
27.63 13.70
STH23L 75 7.9 8.1 4.2 5.2 3.4 3.1
91.84 92.47 90.58 92.00 90.35 87.94
35.27 31.35 41.67 32.97 59.97 62.20
2.6 3.7 7.5
89.52 91.25 94.84
53.80 39.90 8.54
WiscU 75 3.8 4.1 6.0 3.2 6.4 5.1
91.47 93.88 91.82 90.87 92.57 94.15
38.45 16.55 52.73 50.35 19.50 9.91
2.6 3.7 4.8
90.68 92.77 94.67
44.40 27.40 22.20
WiscL 75 7.9 7.0 6.4 5.1 4.5 5.7 5.1 3.8 3.2 3.5 3.2
95.69 92.37 92.99 91.43 93.58 92.41 85.64 86.42 84.50 91.70 92.29
0.09 34.30 45.00 46.85 9.75 40.70 93.85 70.85 79.10 46.03 26.35
2.7 3.7 5.6
87.67 91.78 94.60
67.73 53.13 42.13
USH8U
75 6.0 5.4 6.0 5.7 3.8 4.1
91.56 91.81 92.94 92.31 91.72 93.18
71.9 71.7 40.3 54.3 50.9 37.1
2.7 3.8 4.9
88.33 90.64 92.31
60.37 52.57 23.77
USH8L 75 3.8 4.1 4.8 5.4 6.4 5.7
92.85 93.57 93.08 94.02 94.10 93.88
100.3 49.7 51.6 24.3 35.4 40.8
2.7 3.8 6.1
90.63 92.39 94.64
63.87 39.83 17.20
I894U 100 4.8 4.1 6.4 6.4 4.4 5.1
94.15 93.61 92.35 92.94 87.60 91.37
25.33 56.60 55.70 49.30 97.97 62.83
3.7 4.8 5.9
89.50 93.13 94.89
67.57 50.50 46.83
I894L 100 4.8 7.0 7.0 7.3 9.5 8.3
90.33 92.66 93.20 93.43 94.33 91.14
91.00 41.73 72.27 12.00 0.41 72.13
5.0 6.1 8.3
89.39 91.92 94.02
80.73 79.30 16.37
STH21L
100 7.0 7.6 7.6 7.0
89.86 90.58 90.60 91.86
70.1 39.4 90.8 108.0
2.7 3.7
89.00 92.10
53.77 62.53
90
7.0 5.1
91.38 91.29
105.0 90.3
3.4 Correlations of Lab and Field Results
The analysis included in this study cover density and permeability results. For
each of these two properties, three types of measurements were collected. They include
field measurements taken on the surface of the pavement, laboratory measurements taken
for field cores, and laboratory measurements for specimens produced in the lab using the
SGC. The following sections include the correlations determined between pairs of these
measurements as determined to be relevant to the objectives of the project.
3.4.1 Correlation between Field Density and Lab Density
Figure 3.7 shows the comparison of the density measured in the field by using the
nuclear gauge and the density measured in the lab by using the Corelok device. The
results plotted indicate that there is generally a good relationship between these two
measures. It is also observed that the scatter around the equality line is within the 95%
upper and lower bounds, which is an acceptable range due to inherent variability in field
studies.
91
82
84
86
88
90
92
94
96
98
82 84 86 88 90 92 94 96 98
Fiel
d D
ensi
ty (%
)
Lab Density (%)
Figure 3.7 Relationship between field and lab density
3.4.2 Correlation between Field Permeability and Lab Permeability of Field Cores
As shown in Table 3.1, the laboratory permeability was measured for all field
cores taken from all projects. Figure 3.8 shows the relationship between the field
measurements taken on the surface of the pavement using the NCAT device and the lab
permeability measured on field cores for both fine-graded and coarse-graded mixes.
Although a line of equality was expected in the relationship, it is clear from Figure 3.8
that the field permeability values of most pavements are higher than the lab permeability
by a significant margin ranging between 10 percent and 100 percent. This result can be
92
explained by the fact that water during field permeability testing can flow in lateral
directions, while in the permeability measurement in the lab water can only flow
vertically due to the surrounding membrane.
In the fine-graded mix, despite the wide scatter, and the differences in values,
there exists a relationship (R2 = 0.49) between the lab and the field measurements as
shown in Figure 3.8. This relationship although not very strong, there is a strong trend
and thus can possibly be used as a method for the estimation of the field permeability
based on the lab permeability.
In the case of coarse-graded mixes however, the differences between the lab and
the field permeability are much higher than in the fine-graded mix. In addition, a very
poor trend is found for the relationship between these measurements. This could be
explained by the reason that the coarse-graded mix generates rougher pavement surface
than the fine-graded mix, and then, the gasket seal applied underneath the NCAT field
permeameter cannot be perfectly sealed on the rough surface of coarse mixes. Therefore,
the water could leak along the sides of the sealant gasket resulting in higher, and possibly
inaccurate permeability values.
93
1
10
100
1000
104
105
1 10 100 1000
Fine-graded Coarse-graded
Fiel
d Pe
rmea
bilit
y (x
10-5
cm
/s)
Lab Permeability (x 10-5 cm/s)
Coarse, R2 = 0.06
Fine, R2 = 0.49
Figure 3.8 Relationship between field permeability and lab permeability of coarse-
and fine-graded mixes
These results raised the question of whether the use of NCAT permeameter is a
valuable method to measure permeability in the field. Since it is found that the field
permeability measured by the NCAT device correlates well to the true permeability
measured in the laboratory for fine-graded mix, an attempt was made to differentiate
between coarse- and fine-graded mixes, and to evaluate if the NCAT device is a good
method to measure field permeability of unknown graded mixture.
The percent passing 2.36 mm sieve (P8) was selected as a criteria to identify the
gradation of mixtures. All mixtures used in this study were classified into three main
groups according to the percent passing 2.36 mm sieve as shown in Table 3.3.
94
Table 3.3 Classifications of the mixtures based on the percent passing 2.36 sieve (P8)
Percent passing 2.36 mm sieve Mixture > 45% STH23U, STH23L, WiscU, WiscL,
USH110U, USH110L 40-45% STH21U, STH21L, I894U, I894L,
USH8U, USH8L < 40%
I94, USH20U, USH20L, STH17
Figure 3.9 shows three graphical plots for the relationships between field and lab
permeability sorted based on the aggregate gradation type. For the mixtures containing
P8 higher than 45%, Figure 3.9(a) shows that there is a strong relationship between field
and lab permeability (R2 = 0.80). In Figure 3.9(b), the R2 values reduce to 0.10 as the P8
falls between 40% and 45%, and no correlation was found when the P8 is lower than
40% (Figure 3.9(c)). The exponential functions that best fit the data in three plots are:
P8 > 45%, Field K = 8.34 x e0.055 (Lab K) R2 = 0.80 (3.4)
40%<P8<45%, Field K = 192.48 x e0.01 (Lab K) R2 = 0.10 (3.5)
where field K is the field permeability measured by the NCAT device, and lab K is the
permeability of field cores taken from the field and measured in the laboratory using
ASTM D5084 method.
95
1
10
100
1000
104
1 10 100
Fiel
d Pe
rmea
bilit
y (x
10-5
cm
/s)
Lab Permeability (x 10-5 cm/s)
y = 8.34 e0.055x
R2 = 0.80
a) Mixtures with P8 > 45%
b) Mixtures with 40% < P8 < 45% c) Mixtures with P8 < 40%
Figure 3.9 Relationship between field permeability and lab permeability based on %passing 2.36 mm sieve
Based on the results provided in this study, it was observed that the permeability
measured by the NCAT device gives a good correlation with the permeability measured
10
100
1000
104
1 10 100
Fiel
d Pe
rmea
bilit
y (x
10-5
cm
/s)
Lab Permeability (x 10-5 cm/s)
y = 192.48 e0.01x
R2 = 0.10
10
100
1000
104
105
1 10 100
Fiel
d Pe
rmea
bilit
y (x
10-5
cm
/s)
Lab Permeability (x 10-5 cm/s)
y = 660.14 e0.0099x
R2 = 0.03
96
in the lab, particularly for mixtures with P8 aggregate content higher than 45%. It is
recommended that the measuring an index of permeability with the NCAT device can be:
1) appropriately used for fine-graded mixtures with P8 higher than 45%, 2) used with
caution for fine-graded mixtures with P8 between 40-45%, and 3) not a good method for
measuring the field permeability of mixtures with P8 lower than 40%.
3.4.3 Correlation between Laboratory Permeability of Field Cores and Predicted
Permeability Using Lab Compacted Specimens
It is clear from the previous section that field permeability measured with the
NCAT device cannot be assumed reliable in comparison to the permeability of field cores
tested in the laboratory. However, in the case of fine mixtures there is a good correlation
between field and lab permeability. The remaining question is if the field permeability of
nine mixtures can be predicted from SGC compacted specimen. This question is
important to answer because if the prediction is possible, then it can be included as part
of the mixture design procedure. As indicated earlier in section 3.3, the SGC specimens
were produced according to Method B. Figure 3.10 shows the comparison between the
measured permeability values of field cores and the values predicted based on the SGC
specimens. The best fit regression line, fitted with a zero intercept, has a slope of 0.763,
as shown in the following equation:
Predicted K = 0.763 (Measured K for field cores), R2 = 0.43 (3.6)
It is observed that the measured values are higher than the predicted values,
particularly at higher permeability levels (> 50x10-5 cm/s). Also the slope of regression
line is significantly different from the value of one, or the slope of equality line. The
97
majority of the scatter is, however, close to the equality line, and falls within the 95%
upper and lower bounds from the equality line. Therefore, although the regression trend
does not match the equality, it is believed that this procedure could provide a reasonably
acceptable estimate of the permeability.
These results support are not totally new because others conducting research in
this area have also found similar trends. In a previous study at NCAT, reported by
Cooley (2002), the authors have indicated that the SGC specimens could be used to
estimate the permeability of the specimen compacted in the field.
0
50
100
150
0 50 100 150
Pred
icte
d Pe
rmea
bilit
y (x
10-5
cm
/s)
Measured Permeability (x 10-5 cm/s)
y = 0.763 x
Equality Line
Figure 3.10 Predicted vs. Measured Permeability
98
3.5 Summary of Findings of Laboratory Study
The previous sections included a detailed presentation of all data collected in the
laboratory study. The following points summarize the findings of this study.
1) There is a good relationship between the density measured in the field using the
nuclear gauge and density measured in the lab using the Corelok Device. The
nuclear gauge is therefore considered as an appropriate method to measure in-
place density without damaging the pavement surface.
2) There is a good relationship between the field permeability and the laboratory
permeability measured on field cores of fine-graded mixes with P8 higher than
45%. However, the relationship between the field permeability and the laboratory
permeability measured on field cores of coarse-graded mix (P8 lower than 40%)
is very poor. It should be noted that the NCAT permeability devices, with all its
limitations, could possibly be used in the field (particularly for fine-graded mix
with P8 higher than 45%) to measure an index related to true permeability values
of field cores under well-controlled conditions. The true permeability is defined
here as the values measured using the ASTM standards, which are recognized by
many to be the best practice for granular materials. However, to measure the field
permeability of coarse-graded mix (P8 lower than 40%), an approach to prevent
water leakage along the sealant due to rough pavement surface should be
established. The only other alternative that could be recommended at this time is
to use cores extracted from the pavement and tested in the laboratory.
3) The lab permeability of SGC specimens, produced by Method B (constant number
of gyrations with different sample sizes), provides a good prediction tool for the
99
lab permeability of field cores. Therefore, the use of Method B to produce the
SGC specimen could be a good potential approach for including permeability
criteria in the mixture design process.
CHAPTER FOUR
AIR AND WATER PERMEABILITY STUDY
This chapter presents the results of comparative air and water field permeability
measurement of in-place asphalt pavement layers as well as a study of preferential flow
paths and their relation to field and laboratory water permeability test results.
4.1 Development of Air Permeameter for Asphalt Pavements
In recent years, field permeability testing of in-place asphalt pavement was
commonly performed with water-based, falling head permeameters such as the
previously described NCAT device. Based on literature reviews conducted during the
project proposal and initial study phases, the NCAT device was selected for use and
deployed on initial field studies. Field permeability testing with this device identified a
number of constraints which may inhibit the practical use of this device for field and/or
acceptance testing of in-place pavements:
1. Permeability testing at any selected location is time consuming and labor
intensive. Furthermore, a significant amount of water was necessary to
initiate testing and large amounts of water are required to complete testing,
particularly for coarse graded mixes with high permeability.
100
2. The NCAT permeameter requires an intimate seal between the pavement
surface and the bottom of the device to eliminate water bleeding which
would invalidate test results. This seal proved difficult to achieve and
virtually impossible to verify during testing. Three separate sealing
methods/materials were utilized throughout this study in an effort to
achieve an adequate barrier to water bleed.
3. Repeated testing at selected test locations indicated a significant variation
in permeability measurements, as indicated by changes in the time
required to achieve a desired head change. Initially it was felt that these
variations were the result of variable degrees of saturation during testing.
However, after a number of tests conducted with pre-saturated pavements
it was determined that the changing time intervals were more likely due to
changes if the length of in-pavement flow paths which cannot be measured
non-destructively during testing.
The ROMUS air permeameter was envisioned, designed and constructed during
this project in an effort to eliminate the above constraints while still providing useful field
measurements. The ROMUS device is based on the falling-head air permeameter
principle with one noted exception: a vacuum chamber is used to draw air through the
pavement as opposed to a pressurized chamber forcing air into the pavement. While
fundamentally consistent with air flow measures of earlier devices, the vacuum chamber
also serves to enhance the seal between the device and the pavement surface. This is in
101
contrast to a pressurized chamber which must be ballasted to remain in contact with the
pavement surface.
Figure 4.1 provides a schematic illustration of the ROMUS air permeameter and
Figure 4.2 illustrates the completed device in position for field testing.
Figure 4.1 Schematic Illustration of ROMUS Air Permeameter
102
Figure 4.2 Illustration of the ROMUS Air Permeameter
The main components of the ROMUS air permeameter include a hand operated
grease gun, base seal reservoir, vacuum chamber, automatic vacuum pump and valve,
digital pressure gauge, and digital display. To initiate testing, the bottom of the ROMUS
device is first sealed to the pavement surface by way of a grease seal. The sealant grease
is manually pumped through the device into a recessed base ring which was sized to
replicate the opening of the NCAT water permeameter and designed to eliminate
problems observed with the various sealing techniques used for the NCAT device.
Manually pumping of the grease through the recess ring appears to provide an efficient
seal that can easily conform to the surface irregularities present on asphalt pavements of
the type investigated during this study.
Once the device has been sealed to the pavement surface, pressing of the start
button initiates a fully automated system that first creates a vacuum within the internal
pressure chamber. When the vacuum pressure reaches a value of approximately 25
inches of water (47mm Hg), effectively simulating the maximum head of water used with
the NCAT device, a valve automatically opens to allow air to be drawn through the
pavement layer into the vacuum chamber. A timing system with a resolution of 1
millisecond initiates when the vacuum pressure reaches 24 inches of water and
continually records the time until the internal pressure reaches 8 inches of water. For this
research project, the ROMUS device was programmed to record four timing increments,
103
each representing a change in vacuum pressure equivalent to 4 inches of water. This set-
up simulates a falling head water permeability test with head drops from 24 – 20 inches,
20 – 16 inches, 16 – 12 inches and 12 – 8 inches. Once the test is complete, the four
timing increments are displayed on a digital display for manual recordation. A full test
sequence, including initial vacuum draw and four incremental measurements, can be
completed in less than one minute.
Repeated testing with the ROMUS device indicates consistent results from one
time increment to the next as well as from one test trial to the next. Figures 4.3 and 4.4
illustrate results from two sites located along USH 20. These sites were selected as
representative of trials with both low and high permeability readings. Equivalent water
permeabilities are displayed based on individual recorded time increments (24-20, 20-26,
16-12, 12-8 inches of water) as well as from the overall recorded time (24-8). As shown,
the results are consistent across all test trials for both sites, with an overall coefficient of
variation of 2.2% for the USH20U site and 7.7% for the USH20L site.
104
ROMUS Permeability ResultsUSH20U - Site 5
0102030405060708090
100
1 2 3
Trial Number
Equi
vale
nt W
ater
Perm
eabi
lity
(10
-5 c
m/s
)
24-20 20-18 18-12 12-8 24-8
Figure 4.3 ROMUS Permeability Results for USH20U – Site 5
105
ROMUS Permeability ResultsUSH20L - Site 9
0
1000
2000
3000
4000
5000
1 2 3
Trial Number
Equi
vale
nt W
ater
Perm
eabi
lity
(10
-5 c
m/s
)
24-20 20-18 18-12 12-8 24-8
Figure 4.4 ROMUS Permeability Results for USH20L – Site 9
4.2 Comparison of Field Permeameter Readings
The ROMUS device was used in tandem with the NCAT device during testing on
seven projects incorporating 72 test sites with ranging permeabilities. Figure 4.5
provides an aggregate comparison of equivalent water permeabilities measured by the
ROMUS device versus NCAT water permeability readings. Figure 4.6 provides a
grouped comparison of permeability readings based on gradation classifications
described in Table 3.3.
106
Field Permeability Comparison
All Mixesy = 16.833x0.5734
R2 = 0.5193
1.E+00
1.E+01
1.E+02
1.E+03
1.E+04
1.E+05
1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05
NCAT Water Permeability (x10-5cm/s)
RO
MUS
Equ
iv. W
ater
Per
mea
bilit
y (x
10-5
cm/s
)
Figure 4.5 Aggregate Comparison of Field Permeability Readings
Field Permeability Comparison
Fine P8>45%y = 14.49 x 0.6657
R2 = 0.5490
Fine 40%<P8<45%y = 21.609 x 0.5651
R2 = 0.6676
Coarse P8<40%y = 1.8429 x 0.851
R2 = 0.4344
1.E+00
1.E+01
1.E+02
1.E+03
1.E+04
1.E+05
1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05
NCAT Water Permeability (x10-5cm/s)
RO
MU
S Eq
uiv.
Wat
erPe
rmea
bilit
y (x
10-5
cm/s
)
Coarse P8<40% Fine 40% < P8 < 45% Fine P8>45%
Figure 4.6 Group Comparisons of Field Permeability Readings
107
The data displayed in Figure 4.6 indicates better agreement between permeability
devices for fine graded mixes with P8 > 40%. Figure 4.7 provides a field permeability
comparison for all fine mixes investigated.
Field Permeability Comparison
All Fine Mixesy = 18.317 x 0.5934
R2 = 0.822
1.E+00
1.E+01
1.E+02
1.E+03
1.E+04
1.E+00 1.E+01 1.E+02 1.E+03 1.E+04
NCAT Water Permeability (x10-5cm/s)
ROM
US
Equ
iv. W
ater
Per
mea
bilit
y (x
10-5
cm/s
)
Figure 4.7 Field Permeability Comparison for All Fine Mixes
It is recognized that the data set used in the above comparison plots is limited;
however, it appears that the ROMUS air permeameter may be well suited to serve as an
alternate field testing device for measuring in-place permeabilities of asphalt pavement
layers.
108
4.3 Preferential Flow Path Testing
Field and laboratory permeability measurements of asphalt pavements are directly
related to the number, size, and interconnectivity of void pathways within the test sample.
Correlations between field and laboratory water permeability readings described in
Section 3.4.2 indicated laboratory measures typically produce permeabilities significantly
lower than field obtained values. In an effort to more fully investigate the effects of
preferential void pathways on measured permeabilities, a void pathways indicator was
developed to better quantify the distribution of void pathways in compacted asphalt
layers. This device was developed based on research findings presented by Hall and Ng
(2001) with modifications to provide quantitative, rather than qualitative results on
recovered asphalt cores or gyratory compacted samples.
The void pathways indicator developed during this research effort consists of a
one-inch diameter water standpipe, ballast weights, isolation plates, and collection tubing.
Figure 4.8 provides a schematic illustration of this device and Figure 4.9 illustrates the
device as set up during testing. Tests were conducted on vacuum saturated core
specimens after sandblasting of the perimeter wall was completed to remove any residual
coring smear. For all tests, isolation plates were position at the top and bottom of the
cores to segregate water exiting at the top and perimeter surfaces, respectively, from that
which traveled vertically through the core specimen. For thicker cores in excess of 7 cm,
an addition isolation plate was positioned at the mid-depth of the core to segregate upper
and lower perimeter exit water. Individual core tests typically utilized three fillings of
the standpipe to provide sufficient water collections at each exit location.
109
Figure 4.8 Schematic Illustration of the Void Pathways Indicator
Figure 4.9 Illustration of the Void Pathways Indicator Test Setup
Upper Isolation Plate
Lower Isolation Plate
Water Standpipe
Ballast
Potential Flow Paths
HMA Core
Sample
Core Support
Bottom CollectionPlate
110
Void pathways test results indicate a wide variation in preferential water
pathways. Figures 4.10 and 4.11 illustrate the percentage of exit water which traveled
vertically through the core specimens versus core thickness and core density,
respectively. This parameter is of particular importance during comparisons of field and
laboratory permeability measurements as cores which exhibit preferential vertical water
flow should be more likely to produce comparable results between field permeability
measures, which do not constrain directional water flow, and laboratory measures which
are set up to allow only vertical water flow through core specimens. This statement
assumes, however, that vertical flow out of the in-place asphalt layer is not constrained
by supporting pavement layers. The best-fit data trend illustrated in Figure 4.10 clearly
indicates preferential vertical flow reduces as the core thickness increases. The data
provided in Figure 4.11 shows poor correlations between core density and vertical flow
preference for all but the midrange core thickness where increased density tends to
reduce preferential vertical flow.
111
Void Pathways Comparison
y = 0.2614x2 - 10.648x + 91.904R2 = 0.3691
0
10
20
30
40
50
60
70
80
90
100
0 5 10 15
Core Thickness, cm
% B
otto
m E
xit W
ater
Figure 4.10 Void Pathways Comparison Based on Core Thickness
Void Pathways Comparison
0102030405060708090
100
85 90 95 100
Core Density, %Gmm
% B
otto
m E
xit W
ater
H<6cm 6cm<H<9cm H>9cm
Figure 4.11 Void Pathways Comparison Based on Core Density
112
Figures 4.12 through 4.20 illustrate % bottom exit water versus core thickness,
density and t/NMAS ratio for data segregated by mix gradation, aggregate type and
NMAS. For coarse mixes with P8<40% (Figures 4.12 to 4.14) the % bottom exit water
appears to be correlated only to core thickness with % bottom exit water reducing as
thickness increases. For fine mixes with 40%<P8<45% (Figures 4.15 to 4.17) the %
bottom exit water appears to be influenced by both core thickness and t/NMAS ratio with
the % bottom exit water reducing as core thickness and t/NMAS ratio increase. For the
fine mixes with P8>45% (Figures 4.18 to 4.20) the % bottom exit water appears to be
influenced only by the t/NMAS ratio which reduces % bottom exit water as t/NMAS
increases, particularly for the gravel source mixes.
Void Pathways ComparisonCoarse Mixes (P8<40% )
0
20
40
60
80
100
0 2 4 6 8 10 12 14
Core Thickness, cm
% B
otto
m E
xit W
ater
GR25 LS9.5 LS19 LS25
Figure 4.12 Void Pathways of Coarse Mixes Based on Core Thickness
113
Void Pathways ComparisonCoarse Mixes (P8<40% )
0
20
40
60
80
100
88 89 90 91 92 93 94 95
Core Desnity, %Gmm
% B
otto
m E
xit W
ater
GR25 LS9.5 LS19 LS25
Figure 4.13 Void Pathways of Coarse Mixes Based on Core Density
Void Pathways ComparisonCoarse Mixes (P8<40% )
0
20
40
60
80
100
0 1 2 3 4 5 6 7 8
t/NMAS Ratio
% B
otto
m E
xit W
ater
Gravel Limsetone
Figure 4.14 Void Pathways of Coarse Mixes Based on t/NMAS Ratio
114
Void Pathways ComparisonFine Mixes (40% <P8<45% )
0
20
40
60
80
100
0 2 4 6 8 10
Core Thickness, cm
% B
otto
m E
xit W
ater
GR12.5 GR19 LS12.5 LS19
Figure 4.15 Void Pathways of Midrange Fine Mixes Based on Core Thickness
Void Pathways ComparisonFine Mixes (40% <P8<45% )
0
20
40
60
80
100
89 90 91 92 93 94 95
Core Density, cm
% B
otto
m E
xit W
ater
GR12.5 GR19 LS12.5 LS19
Figure 4.16 Void Pathways of Midrange Fine Mixes Based on Core Density
115
Void Pathways ComparisonFine Mixes (40% <P8<45% )
0
20
40
60
80
100
0 1 2 3 4 5 6
t/NMAS Ratio
% B
otto
m E
xit W
ater
Gravel Limestone
Figure 4.17 Void Pathways of Midrange Fine Mixes Based on t/NMAS Ratio
Void Pathways ComparisonFine Mixes (P8>45% )
0
20
40
60
80
100
0 2 4 6 8 10
Core Thickness, cm
% B
otto
m E
xit W
ater
GR12.5 GR19 LS12.5 LS19
Figure 4.18: Void Pathways of Fine Mixes Based on Core Thickness
116
Void Pathways ComparisonFine Mixes (P8>45% )
0
20
40
60
80
100
90 91 92 93 94 95 96
Core Density, %Gmm
% B
otto
m E
xit W
ater
GR12.5 GR19 LS12.5 LS19
Figure 4.19 Void Pathways of Fine Mixes Based on Core Density
Void Pathways ComparisonFine Mixes (P8>45% )
0
20
40
60
80
100
0 1 2 3 4 5 6 7 8
t/NMAS Ratio
% B
otto
m E
xit W
ater
Gravel Limestone
Figure 4.20 Void Pathways of Fine Mixes Based on t/NMAS Ratio
117
The impact of preferential vertical pathways on comparative field and laboratory
water permeability measures was also examined. Figure 4.21 illustrates a comparison of
measured water permeability ratios, calculated as the ratio of field permeability to lab
permeability, versus % bottom exit water for all cores tested. On this aggregate level, no
discernable trend is evidenced.
Water Permeability Comparison
0.1
1.0
10.0
100.0
1000.0
10000.0
0 20 40 60 80 100
% Bottom Exit Water
NCAT
Fie
ld -
UW L
abW
ater
Per
mea
bilit
y Ra
tio
Figure 4.21 Water Permeability Comparison Based on Void Pathways
Figures 4.22 to 4.24 illustrate water permeability ratios versus % bottom exit
water for mix types segregated by the P8 percentage. For the coarse (P8<40%) and
midrange fine (40%<P8<45%) gradations, the water permeability ratios tend to decrease
as the % bottom exit water increases (Figures 4.22 and 4.23). For the coarse mixes, the
best-fit trend line suggests that even for cores with a high degree of vertical flow
preference (% bottom exit water > 80%) field and lab permeability values may differ by
an order of magnitude. In contrast, the trend line for the midrange fine gradations
118
indicates better agreement between field and lab permeability measures as preferential
vertical flow increases.
For fine mixes with P8>45%, the water permeability ratio does not appear to be
affected by preferential vertical pathways as the available data indicates permeability
ratios near unity for all cores examined.
Water Permeability Comparison
y = 177.43x-0.5745
R2 = 0.2831
1.0
10.0
100.0
1000.0
10000.0
0 20 40 60 80 100
% Bottom Exit Water
NC
AT
Fiel
d - U
W L
abW
ater
Per
mea
bilit
y R
atio
Coarse P8<40%
Figure 4.22 Water Permeability Comparison for Coarse Mixes
119
Water Permeability Comparison
y = 419.75x-1.1817
R2 = 0.6598
1.0
10.0
100.0
1000.0
0 20 40 60 80 100
% Bottom Exit Water
NC
AT
Fiel
d - U
W L
abW
ater
Per
mea
bilit
y R
atio
Fine 40%<P8<45%
Figure 4.23 Water Permeability Comparison for Midrange Fine Mixes
Water Permeability Comparison
y = 0.0029x + 1.256R2 = 0.0033
0.0
1.0
2.0
3.0
4.0
5.0
0 20 40 60 80 100
% Bottom Exit Water
NC
AT
Fiel
d - U
W L
abW
ater
Per
mea
bilit
y R
atio
Fine P8>45%
Figure 4.24 Water Permeability Comparison for Fine Mixes
120
CHAPTER FIVE
CONCLUSIONS AND RECOMMENDATIONS
5.1 Summary of Findings
5.1.1 Field Study
Based on an analysis of the data collected in the field study, the following findings
can be stated:
1) Density and permeability characteristics of Superpave mixes are based on project-
specific variables. Base type, source, gradation, and Ndes level all influence field
density and permeability. No discernible trend was observed between density and
permeability for coarse-graded mixes.
2) A clear relationship between layer thickness and permeability was not established.
Layer thickness was a factor on a project-specific basis, with some projects
indicating it was significant, while others found it not significant.
3) Fine-graded limestone-source mixes compacted on PCC, and those designed at a
higher Ndes level, were more permeable than other mixes produced from different
sources or constructed on different subsurface layers.
4) For fine-graded mixes, the t/NMAS ratio showed an influence on achieving
density, particularly below a ratio of 2 for gravel-source mixes and a ratio of 3 for
limestone-source mixes. For limestone-source mixes outside the current WisDOT
t/NMAS range of 3 to 5, it was more difficult to achieve density below a ratio of
3, and possible to achieve a 92% density above a ratio of 5. No clear relationship
was found between t/NMAS ratios and permeability.
121
5) For coarse-graded mixes, mixes compacted at smaller t/NMAS ratios for
limestone-source were more permeable than higher ratios, but no trend was
observed for the gravel-source mix.
6) The factors that affected density growth during compaction included mat
temperature, number of passes, and their interaction (a declining mat temperature
occurs with more passes).
7) It is found that gradation of the aggregate could be linked to permeability. The
ratio of (%P1/2 - %P3/8) / (%PNo.4-%PNo.8) had a good correlation with
permeability with high ratios showing lower permeability. In addition, higher
permeability was measured as the gaps increase between the coarse aggregates
(%P1/2” and %P3/8”) and/or the fine aggregates (%P4 and %P8). This suggests
that relative differences in these sieves may have an effect on internal void
structure, and thus measured permeability, of the compacted material. This trend
could be used in mix design by controlling the ratio to limit permeability by either
reducing the difference between the coarse sieves, fine sieves, or both.
8) The air permeameter produced results which were comparable to those obtained
with the NCAT water permeameter, particularly for the fine-graded mixes. The
initial results show that the air permeameter produces time efficient, reproducible
results and appears to be a viable alternative for the NCAT water permeameter.
5.1.2 Laboratory Study
Based on the analysis of data collected in the laboratory study, the following findings
can be stated:
122
1) A good relationship exists between the density measured by using the nuclear
gauge and the density measured in the lab using the Corelok Device. The nuclear
gauge has therefore been found to be a rapid, reliable, and non-destructive method
to accurately measure in-place density of asphalt mixtures in the field.
2) The NCAT field permeability device was found to give results that sometimes
compares well to laboratory measurements done on field cores but not always.
For fine-graded mixture with P8 higher than 45%, field permeability measured by
the NCAT device strongly correlates to laboratory permeability measured on field
cores taken from same pavements section. However, the relationship is not one to
one ratio. The field permeability values could be approximately an order of
magnitude higher than the lab permeability. This could be explained by the
multiple flow directions in the field permeability measurement. The coefficients
of correlation for the mathematical relationship found is high (R2 = 0.80). This
indicates that the NCAT permeability devices, with all its limitations, could be
used in the field for fine-graded mixture (with P8 > 45%) to measure an index of
permeability reliability. The measured values can then be related to true
permeability of field cores measured by the ASTM D5084 conducted under well-
controlled conditions. There is a concern, however, in using the NCAT device for
measuring the field permeability of mixtures with P8 lower than 40%, since very
poor correlation was found for the relationship between field and lab permeability
in this study. The modification of NCAT device is therefore necessary in order to
prevent water leakage along the sealant due to rough pavement surface,
particularly for mixtures with coarse gradation.
123
3) A method was developed to compact specimens in the SGC at various sample
sizes that could be used to estimate relatively well the permeability of the
specimens taken out from pavements in the field. The permeability measured on
these SGC specimens correlates to the permeability measured on the field cores
with a relationship of one to one. Therefore, this method (called in the report
Method B) could be used for predicting the permeability of asphalt mixtures in
the field. If this method can be validated, then the permeability can be included as
a design requirement.
4) A method, and related equipment, were developed for quantifying the preferential
void pathways in compacted asphalt layers. The degree of vertically connected
void pathways was found to be best correlated to the pavement layer thickness,
with greater thicknesses producing a reduction in preferential vertical void
pathways. Correlations between field/lab water permeability ratios and
preferential vertical void pathways indicate that field and laboratory permeability
values can only be expected to be in near agreement when the degree of
preferential vertical void pathways exceeds 80% for fine mixes. For coarse mixes
with a high degree of preferential vertical void pathways, field/lab water
permeability ratios of 10 or more may be expected.
5.2 Recommendations
5.2.1 Guidelines for Selection of Pavement Thickness in Wisconsin
It is recommended that no changes be made to the current layer thickness values
and t/NMAS ratios in the specifications. Density and permeability characteristics of
124
Superpave mixes are found to depend on several project-specific variables, such as base
type, source, gradation, Ndes level, layer thickness, and t/NMAS ratio. No compelling
evidence is found in the data to alter layer thickness and t/NMAS ratios, without
accounting for the other remaining project-specific variables. It is however important to
recognize that the current recommendations do not ensure achieving density nor limit
permeability. Difficulty in achieving density or exceeding acceptable permeability is
influenced by several interacting factors.
5.2.2 Recommendations on Laboratory and Field Permeability Testing Procedure
To measure laboratory permeability and estimate field permeability, the following
testing procedure is recommended for the mixture design and quality control of
Superpave mixtures in Wisconsin. Figure 5.1 shows the detail steps required for the
procedure using samples compacted in the laboratory.
1) Evaluate the job mix formula to determine if the mix is fine or coarse graded and
determine the percent passing the No. 8 sieve.
2) For fine graded mixes with P8>45%, produce Superpave gyratory compacted
(SGC) specimens from representative loose materials according to Method B
described in this report.
3) Measure the true permeability of the SGC specimens in the laboratory according
to ASTM D5084.
4) Estimate the lab and field permeability of the field compacted specimens based on
the target field density using the relationship between lab permeability and
density derived from testing the samples with different sizes.
125
5) For fine-graded mix with P8 < 45%, and for coarse-graded mix, the lab
permeability cannot be used to estimate field permeability because there is no
correlation between the field permeability, which includes flow in multiple
directions, and the one-dimensional permeability measured in the lab on a sample
compacted in the SGC. To estimate the true one-dimensional permeability a core
should be extracted and used for measuring permeability in the lab.
6) For quality control purposes of fine graded mixes with (P8 > 45%), the NCAT
device can be used to measure an index of permeability in the field. The actual
(one dimensional) permeability can be predicted using the following equation.
The estimated value can be compared to the design permeability value.
• For fine-graded mix (P8 > 45%)
Field K = 8.34 x e0.055 (Lab K)
For other types of mixtures, the NCAT device can be used as a quality control.
126
Obtain Loose Mix Based on JMF
Predict Lab Permeability (ASTM D 5084) of Field Compacted Specimens
Fine-Graded Mix P8 > 45%
Produce SGC Specimens Using Method B
Measure Lab Permeability (ASTM D 5084)
Fine-Graded Mix P8 < 45% and
Coarse-Graded Mix
Predict Field Permeability (NCAT Device) based on Lab Permeability (ASTM D 5084)
Prediction cannot be made, and NCAT device is not recommended
Figure 5.1 Procedures for designing and predicting field permeability
127
5.2.3 Recommendations for Permeability and Density Criteria for Superpave Mix
Designs in Wisconsin
It is recommended that target permeability and density values ultimately be
established from in-service pavements with recorded performance histories. One such
group of pavements includes accepted warranty projects that have been in service for 5 or
more years. Field permeability and density measures on these pavements can aid in the
development of acceptance values that correlate to good performance.
Until a performance-based determination is made, an interim approach is
recommended that establishes the minimum acceptable density based on median
permeability values. Based on research data included in this report for fine-graded
Superpave mixes, a specified minimum density of 93.8% would be required to limit
permeability to 150x10-5 cm/sec. For coarse-graded Superpave mixes, the research data
does not support the establishment of minimum acceptable densities to control
permeability because of the lack of a unified relationship between density and
permeability that is independent of source or garadtion of mixtures. The limit should
remain at 150x10-5 cm/sec but should be measured directly on a core recovered from
pavement section.
128
REFERENCES
- Al-Omari, A., Tashman, L., Masad, E., Cooley, A., and Harman, T. “Proposed
Methodology for Predicting HMA Permeability,” Journal of the Association of
Asphalt Paving Technologists, Proceedings of the Technical Sessions, Volume 71,
2002.
- ASTM D 5084 “Measurement of Hydraulic Conductivity of Saturated Porous
Materials using a Flexible Wall Permeameter,” Annual Book of ASTM Standards,
American Society for Testing and Materials, 2000.
- ASTM PS 129-01 “Standard Provisional Test Method for Measurement of
Permeability of Bituminous Paving Mixtures Using a Flexible Wall Permeameter,”
Annual Book of ASTM Standards, American Society for Testing and Materials, 2001.
- Bishop, A. and Eldin, G. “Undrained Triaxial Tests on Saturated Sands and Their
Significance in the General Theory of Shear Strength,” Geotechnique, Vol.2, No.1,
1950.
- Choubane, B., Page, G., and Musselman, J. “Investigation of Water Permeability of
Coarse Graded Superpave Pavements,” Research Report FL/DOT/SMO/97-416,
Florida Department of Transportation, Tallahassee, Florida, July 1997.
- Collins, R. Flow of Fluids Through Porous Materials. Reinhold Publishing Corp.,
New York, 1961.
- Cooley, Jr., L.A., “Permeability of Superpave Mixtures: Evaluation of Field
Permeameters”, Final Report, NCAT, 1998.
- Cooley, L. A., Brown, E., and Maghsoodloo, S. “Development of Critical Field
Permeabiltiy and Pavement Density Values for Coarse-Graded Superpave
129
Pavements,” NCAT Report No. 01-03, National Center for Asphalt Technology,
Auburn, AL, September 2001.
- Cooley, L.A., Prowell, B.D., and Brown, E.R. “Issues Pertaining to the Permeability
Characteristics of Coarse-Graded Superpave Mixes,” Journal of the Association of
Asphalt Paving Technologists, Proceedings of the Technical Sessions, Volume 71,
2002.
- Daniel, D.E. “A Note on Falling-Headwater and Rising-Tailwater permeability
Tests,” Geotechnical Testing Journal, Vol.12, No.4, 1989.
- Daniel, D. “State-of-the-Art: Laboratory Hydraulic Conductivity Tests for Saturated
Soils,” In Hydraulic Conductivity and Waste Contaminant Transport in Soil, ASTM
STP 1142, American Society for Testing and Materials, Philadelphia, PA. 1994.
- Hall, K.D., Cruz, J., and Ng, H. “Effects of Testing Time and Confining Pressure on
Falling-Head Permeability Tests of Hot-Mix Asphalt Concrete,” Transportation
Research Record 1723, 2000.
- Hall, K.D., and Ng, H.G. “Development of a Void Pathway Test for Investigating
Void Interconnectivity in Compacted Hot-Mix Asphalt Concrete,” Transportation
Research Record, 1767, 2001.
- Hillel, D. Environmental Soil Physics. Academic Press, San Diego, 1998.
- Huang, B., Mohammed, L., Raghavendra, A., and Abadie, C. “Fundamentals of
Permeability in Asphalt Mixtures,” Journal of the Association of Asphalt Paving
Technologists, Volume 68, 1999.
- James, J.M. “Asphalt Mix Permeability,” FHWA/AR-88/003, September 1965.
130
- Kanitpong, K., Benson, C.H., and Bahia, H.U. “Hydraulic Conductivity
(Permeability) of Laboratory Compacted Asphalt Mixtures,” Transportation Research
Record, 1767, 2001.
- Kanitpong, K., Bahia, H.U., Benson, C.H., and Wang, X. “Effect of Lift Thickness
and Flow Direction on Hydraulic Conductivity (Permeability) of Laboratory
Compacted Asphalt Mixtures,” Presented at the 81st Annual Meeting of the
Transportation Research Board (TRB Committee A2D02 Meeting), Washington,
D.C., 2002.
- Mallick, R.B., Cooley, L.A., Teto, M.R., Bradbury, R.L., and Peabody, D. “An
Evaluation of Factors Affecting Permeability of Superpave Designed Pavements,”
prepared for 2001 Annual Meeting of the Transportation Research Board, 2001.
- McLaughlin, J.F. and Goetz, W.H. “Permeability, Void Content, and Durability of
Bituminous Concrete,” Proceedings, Highway Research Board, Vol. 34, 1955.
- Menard, J.P. “Comparative Analysis of Field Permeability Testing of Compacted
Hot-Mix Asphalt Pavements Using Air and Water Permeameters,” Master
Dissertation, Marquette University, Milwaukee, Wisconsin, 2003.
- Moore, P.J. “Determination of Permeability Anisotropy in a Two-Way
Permeameter,” Geotechnical Testing Journal, Vol. 2(3), pp. 167-169, 1979.
- Paye, B. “Minimum Pavement Thickness for Superpave Mixtures,” Master
Dissertation, Univeristy of Wisconsin-Madison, 2001.
- Vallerga, B.A. and Hicks, R.G. “Water Permeability of Asphalt Concrete Specimens
using Back-Pressure Saturation,” Journal of Materials, Vol. 3, No. 1, 1968.
131
- Vavrik, W.R., Pine, W.J. and Carpenter, S.H. “Aggregate Blending for Asphalt Mix
Design: “The Bailey Method”, Proceedings of the Transportation Research Board
81st Annual Meeting, Washington, DC, January 2002.
- Weaver, A. “Determination of Permeability of Granular Soil By Air Subjected to a
Decreasing Pressure Differential,” Symposium on Permeability of Soils: ASTM
Technical Publication No. 163. American Society for Testing Materials, Philadelphia,
1955.
- Westerman, J. R. “AHTD’s Experience with Superpave Pavement Permeability,”
Presented at the Arkansas Superpave Symposium, Little Rock, Arkansas, 1998.
- Zube, E. “Compaction Studies of Asphalt Concrete pavements as related to the
Water Permeability Test,” Highway Research Board, Bulletin 358, 1962.
- Skempton, A. “The Pore Pressure Coefficients A and B,” Geotechnique, Vol.4, No.1,
1954.
132
Appendix A
Project-Specific Permeability Significance Testing
133
Table A.1 ANOVA Results of Project Significance Testing (Fine Mixes)
I-43 STH-23 STH-23 USH-18 USH-18 USH-10 USH-10
19-mm 19-mm 12.5-mm 19-mm 12.5-mm 19-mm 12.5-mm
Limestone Gravel Gravel Limestone Limestone Limestone Limestone
Variable Coarse Fine Fine Fine Fine Fine Fine
(1) (2) (3) (4) (5) (6) (7) (8)
Degrees of Freedom 14 22 11 31 17 20 13
Thick *** *** *** N/S N/S *** N/SDensity N/S *** ** *** *** * *Thick*Density N/S *** N/S N/S N/S N/S N/S
Testing Variability, % 29 5 23 70 56 37 67
Significance Levels: N/S = Not Significant; * = 0.05 < p-value < 0.10 ; ** = 0.01 < p-value< 0.05; *** = p-value < 0.01
Table A.2 ANOVA Results of Project Significance Testing (Fine Mixes)
I-894 I-894 STH-21 STH-21 USH-8 USH-8 USH-8 USH-8
19-mm 12.5-mm 19-mm 12.5-mm 19-mm #1 19-mm #2 19-mm
pooled 12.5-mm
Limestone Limestone Limestone Limestone Gravel Gravel Gravel Gravel
Variable Fine Fine Fine Fine Fine Fine Fine Fine
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Degrees of Freedom 16 11 16 22 16 18 35 17
Thick N/S N/S *** *** *** * N/S ***Density *** *** *** *** N/S N/S N/S N/SThick*Density N/S ** ** *** N/S N/S N/S N/STesting Variability, % 15 2 5 8 41 82 87 51
Significance Levels: N/S = Not Significant; * = 0.05 < p-value < 0.10 ; ** = 0.01 < p-value< 0.05; *** = p-value < 0.01
134
Table A.3 ANOVA Results of Project Significance Testing (Coarse Mixes)
I-94 USH-20 USH-20 STH-17
25-mm 19-mm 9.5-mm 25-mm
Limestone Limestone Limestone Gravel
Variable Coarse Coarse Coarse Coarse
(1) (2) (3) (4) (5)
Degrees of Freedom 15 11 11 15
Thick N/S *** *** ** Density N/S *** * *** Thick*Density N/S *** N/S ** Testing Variability, % 73 1 2 13
Significance Levels: N/S = Not Significant; * = 0.05 < p-value < 0.10 ;
** = 0.01 < p-value< 0.05; *** = p-value < 0.01
135
Appendix B
Density Growth Plots
136
STH-21 Lower Layer Density Growth
75
77
79
81
83
85
87
89
91
93
95
0 5 10 15 20 25
Roller Passes, n
Den
sity
, %
2in (8)
2in (9)
2.5in
2.75in (4)
2.75in 6)
3in
B.1 Density Growth on STH-21 19-mm Lower Layer Mix
STH-21 Intermediate Layer Density Growth
70
75
80
85
90
95
0 5 10 15
Roller Passes, n
Den
sity
, % 1.625in1.75in2in2.125in2.25in2.375in2.5in
137
B.2 Density Growth on STH-21 12.5-mm Lower Layer Mix
138
USH-8 Lower Layer Density Growth
77
79
81
83
85
87
89
91
93
95
0 2 4 6 8 10
Roller Passes, n
Den
sity
, % 1.25in
1.5in
3in
3.25in
3.375in
3.75in
3.875in
B.3 Density Growth on USH-8 19-mm Lower Layer Mix
USH-8 Intermediate Layer Density Growth
80
82
84
86
88
90
92
94
96
0 5 10 15
Roller Passes,n
Den
sity
, % 1.5in
1.625in
1.75in
1.875in
2.125in
2.5in
2.25in
B.4 Density Growth on USH-8 19-mm Intermediate Layer Mix
139
USH-8 Surface Layer Density Growth
80
82
84
86
88
90
92
94
0 2 4 6 8
Roller Passes, n
Den
sity
, %
1.5in1.5in1.625in2in2.125in2.25in2.375in
B.5 Density Growth on USH-8 12.5-mm Surface Layer Mix
I-94 Lower Layer Density Growth
70
75
80
85
90
95
0 5 10 15 20 25
Roller Passes, n
Den
sity
, % 3.75in4in4in4.625in4.75in4.75in
140
B.6 Density Growth on I-94 25-mm Lower Layer Mix
STH-17 Lower Layer Density Growth
65
70
75
80
85
90
95
0 5 10 15
Roller Passes, n
Den
sity
, %
3.75in3.625in3.75in4.375in4.625in3.75in
B.7 Density Growth on STH-17 25-mm Lower Layer Mix